1. Introduction

The glossaries package is a flexible package, but it’s also a heavy-weight package that uses a lot of resources. As package developer, I’m caught between those users who complain about the drawbacks of a heavy-weight package with a large user manual and those users who want more features (which necessarily adds to the package weight and manual size).

The glossaries-extra package is an attempt to provide a compromise for this conflict. Version 4.22 of the glossaries package is the last version to incorporate new features.^1.1 Future versions of glossaries will just be bug fixes. New features will instead be added to glossaries-extra. This means that the base glossaries package won’t increase in terms of package loading time and allocation of resources, but those users who do want extra features available will have more of a chance of getting their feature requests accepted.

Top

1.1 Package Defaults

I’m not happy with some of the default settings assumed by the glossaries package, and, judging from code I’ve seen, other users also seem unhappy with them, as certain package options are often used in questions posted on various sites. I can’t change the default behaviour of glossaries as it would break backward compatibility, but since glossaries-extra is a separate package, I have decided to implement some of these commonly-used options by default. You can switch them back if they’re not appropriate.

The new defaults are:

• toc=true (add the glossaries to the table of contents). Use toc=false to switch this back off.
• nopostdot=true (suppress the terminating full stop after the description in the glossary). Use nopostdot=false or just postdot to restore the terminating full stop (period).
• noredefwarn=true (suppress the warnings that occur when the theglossary environment and \printglossary are redefined while glossaries is loading). To restore the warnings, use noredefwarn=false. Note that this won’t have any effect if the glossaries package has already been loaded before you use the glossaries-extra package.
• If babel has been loaded, the translate=babel option is switched on. To revert to using the translator interface, use translate=true. There is no change to the default if babel hasn’t been loaded.
• The default style used by \newacronym is short-nolong. (That is, the long form is not shown on first use.)

The examples below illustrate the difference in explicit package options between glossaries and glossaries-extra. There may be other differences resulting from modifications to commands provided by glossaries (see §2 Modifications to Existing Commands and Styles).

1. Basic defaults:

        \documentclass{article}
        \usepackage{glossaries-extra}
   

   This is like:

        \documentclass{article}
        \usepackage[toc,nopostdot]{glossaries}
        \usepackage{glossaries-extra}
   

2. Language defaults:

        \documentclass[british]{article}
        \usepackage{babel}
        \usepackage{glossaries-extra}
   

   This is like:

        \documentclass[british]{article}
        \usepackage{babel}
        \usepackage[toc,nopostdot,translate=babel]{glossaries}
        \usepackage{glossaries-extra}
   

3. Combined with memoir

        \documentclass{memoir}
        \usepackage{glossaries-extra}
   

   This is like:

        \documentclass{memoir}
        \usepackage[toc,nopostdot,noredefwarn]{glossaries}
        \usepackage{glossaries-extra}
   

   However

        \documentclass{memoir}
        \usepackage{glossaries}
        \usepackage{glossaries-extra}
   

   This is like:

        \documentclass{memoir}
        \usepackage[toc,nopostdot]{glossaries}
        \usepackage{glossaries-extra}
   

   Since by the time glossaries-extra has been loaded, glossaries has already redefined memoir’s glossary-related commands.

4. Abbreviations are defined with \newabbreviation:

        \usepackage{glossaries-extra}
        \newabbreviation{svm}{SVM}{support vector machine}
        \begin{document}
        First use: \gls{svm}. Explicit full form: \glsxtrfull{svm}.
        \end{document}
   

   This is the closest to:

        \usepackage{glossaries}
        \newacronym{svm}{SVM}{support vector machine}
        \begin{document}
        First use: \gls{svm}. Explicit full form: \acrfull{svm}.
        \end{document}
   

   If you want to continue using \newacronym then you will need to change the style for the acronym category:

        \usepackage{glossaries-extra}
        \setabbreviationstyle[acronym]{long-short}
        \newacronym{svm}{SVM}{support vector machine}
        \begin{document}
        First use: \gls{svm}. Explicit full form: \glsxtrfull{svm}.
        \end{document}
   

Another noticeable change is that by default \printglossary will now display information text in the document if the external glossary file doesn’t exist. This is explanatory text to help new users who can’t work out what to do next to complete the document build. Once the document is set up correctly and the external files have been generated, this text will disappear.

This change is mostly likely to be noticed by users with one or more redundant empty glossaries who ignore transcript messages, explicitly use makeindex/xindy on just the non-empty glossary (or glossaries) and use the iterative \printglossaries command instead of \printglossary. For example, consider the following:

\documentclass{article}
\usepackage[acronym]{glossaries}
\makeglossaries
\newacronym{laser}{laser}{light amplification by stimulated
emission of radiation}
\begin{document}
\gls{laser}
\printglossaries
\end{document}

The above document will only display the list of acronyms at the place where \printglossaries occurs. However it will also attempt to input the .gls file associated with the main glossary.

If you use makeglossaries, you’ll get the warning message:

Warning: File 'test.glo' is empty.
Have you used any entries defined in glossary 'main'?
Remember to use package option 'nomain' if you
don't want to use the main glossary.

(where the original file is called test.tex) but if you simply call makeindex directly to generate the .acr file (without attempting to create the .gls file) then the transcript file will always contain the message:

No file test.gls.

This doesn’t occur with makeglossaries as it will create the .gls file containing the single command \null.

If you simply change from glossaries to glossaries-extra in this document, you’ll find a change in the resulting PDF if you don’t use makeglossaries and you only generate the .acr file with makeindex.

The transcript file will still contain the message about the missing .gls, but now you’ll also see information in the actual PDF document. The simplest remedy is to follow the advice inserted into the document at that point, which is to add the nomain package option:

\documentclass{article}
\usepackage[nomain,acronym,postdot]{glossaries-extra}
\makeglossaries
\setabbreviationstyle[acronym]{long-short}
\newacronym{laser}{laser}{light amplification by stimulated
emission of radiation}
\begin{document}
\gls{laser}
\printglossaries
\end{document}

(Note the need to set the acronym style using \setabbreviationstyle before \newacronym. See §4 Abbreviations for further details.)

Top

1.2 New or Modified Package Options

If you haven’t already loaded glossaries, you can use any of the package options provided by glossaries when you load glossaries-extra and they will automatically be passed to glossaries (which glossaries-extra will load). If glossaries has already been loaded, then those options will be passed to \setupglossaries, but remember that not all of the glossaries package options may be used in that command.

The new and modified options provided by glossaries-extra are described below:

debug

The glossaries package has a debug option that allows the values false, true and showtargets. The debug=showtargets option was introduced to glossaries v4.32, so if you want to use this option with glossaries-extra you must make sure that your version of glossaries supports it.

The glossaries-extra package extends this option to provide the additional values debug=showwrgloss and debug=all.

The debug=showwrgloss option implements debug=true and uses

---

\glsxtrwrglossmark

---

to show a mark ⋅ just before the write operation performed by the indexing commands. If you use record=hybrid there will be a mark for the write operation to the .aux file for bib2gls and a mark for the write operation to the associated glossary file for makeindex or xindy.

The debug=all option implements both debug=showtargets and debug=showwrgloss.

showtargets

This implements debug=showtargets and can also adjust the way the targets are shown. By default, any links that occur in TeX’s “outer” mode are placed in the margin according to \glsshowtargetouter (provided by the base glossaries package). This can lead to the “too many floats” error if there are multiple instances of commands like \gls in a single line. This option provides a way of changing this to an in-line annotation. Available values:

left

A marker is placed to the left of the link/target and a marginal note is used in outer mode;

right

A marker is placed to the right of the link/target and a marginal note is used in outer mode;

innerleft

A marker and annotation are placed to the left of the link/target in all modes;

innerright

A marker and annotation are placed to the right of the link/target in all modes;

annoteleft

Markers are placed on either side of the link/target and the annotation is on the left;

annoteright

Markers are placed on either side of the link/target and the annotation is on the right.

postdot

(New to version 1.12.) This option is just a shortcut for nopostdot=false.

postpunc

(New to version 1.21.) This option sets the post-description punctuation to the given value. For example: postpunc=; does

     \renewcommand{\glspostdescription}{;}

The value may also be one of the following keywords:

comma:

postpunc=comma is equivalent to

         \renewcommand{\glspostdescription}{,}

dot:

postpunc=dot is equivalent to

         \renewcommand{\glspostdescription}{.\spacefactor\sfcode`\. }

none:

postpunc=none is equivalent to

         \renewcommand{\glspostdescription}{}

The default definition is

     \newcommand*{\glspostdescription}{%
       \ifglsnopostdot\else.\spacefactor\sfcode`\. \fi
     }

where the conditional is determined by the nopostdot package option. The postpunc option removes the conditional from the definition of \glspostdescription. The package options nopostdot and postdot will restore the original definition of \glspostdescription.

The glossaries-extra-stylemods package adjusts the predefined styles that had a hard-coded \space before the number list so that they use \glsxtrprelocation instead (which is defined to \space). You can therefore redefine this command in combination with postpunc to alter the separator before the number list. For example, to have a comma followed by \hfil:

     \usepackage[postpunc=comma,stylemods]{glossaries-extra}
     \renewcommand{\glsxtrprelocation}{\hfil}

Be careful with doing this as it will look odd if the number list is missing. (With bib2gls you can instead redefine \glsxtrprelocation to do nothing and set the location prefixes with loc-prefix which will only apply if the entry has a number list.)

prefix

Load the glossaries-prefix package (if not already loaded).

accsupp

Load the glossaries-accsupp package (if not already loaded).

The glossaries-accsupp package is still experimental and so accessibility features are liable to change.

If you want to define styles that can interface with the accessibility support provided by glossaries-accsupp use the \glsaccess⟨xxx⟩ type of commands instead of \glsentry⟨xxx⟩ (for example, \glsaccesstext instead of \glsentrytext). If glossaries-accsupp hasn’t been loaded those commands are equivalent (for example, \glsaccesstext just does \glsentrytext) but if it has been loaded, then the \glsaccess⟨xxx⟩ commands will add the accessibility information. (See §13.2 Accessibility Support for further details.)

Note that the accsupp option can only be used as a package option (and can’t be set with \glossariesextrasetup) since the glossaries-accsupp package must be loaded before glossaries-extra if it’s required.

stylemods

This is a ⟨key⟩=⟨value⟩ option used to load the glossaries-extra-stylemods package. The value may be a comma-separated list of options to pass to that package. (Remember to group ⟨value⟩ if it contains any commas.) The value may be omitted if no options need to be passed. See §2.11 Glossary Style Modifications for further details. There are two special keyword values: stylemods=default (equivalent to omitting the value) and stylemods=all, which loads all the predefined styles.

undefaction

This is a ⟨key⟩=⟨value⟩ option, which has two allowed values: warn and error. This indicates what to do if an undefined glossary entry is referenced. The default behaviour is undefaction=error, which produces an error message (the default for glossaries). You can switch this to a warning message (and ?? appearing in the text) with undefaction=warn.

Undefined entries can’t be picked up by any commands that iterate over a glossary list. This includes \forglsentries and \glsaddall.

Note that \ifglsused will display ?? in the document text with undefaction=warn if the entry hasn’t been defined, as the underlying boolean variable doesn’t exist and so is neither true nor false. (There will also be a warning in the transcript.) See §2.6 First Use Flag for further details.

indexcrossrefs

This is a boolean option. If true, this will automatically index any cross-referenced entries that haven’t been marked as used at the end of the document. Note that this necessarily adds to the overall document build time, especially if you have defined a large number of entries, so this defaults to false, but it will be automatically switched on if you use the see or seealso keys in any entries (unless autoseeindex=false). To force it off, even if you use the see or seealso key, you need to explicitly set indexcrossrefs to false.

Note that bib2gls can automatically find dependent entries when it parses the .bib source file. The record option automatically implements indexcrossrefs=false.

autoseeindex

(New to v1.16.) This is a boolean option. If true (default), this makes the see and seealso keys automatically index the cross-reference when an entry is defined. If false, the value of those keys will still be stored in their corresponding fields (and can be accessed using commands like \glsxtrusesee and \glsxtruseseealso) but cross-reference won’t be automatically indexed.

Note that the record=only option automatically implements autoseeindex=false.

For example, if an entry is defined as

     \newglossaryentry{foo}{name={foo},description={},see={bar,baz}}

then with autoseeindex=true, this is equivalent to

     \newglossaryentry{foo}{name={foo},description={}}
     \glssee{foo}{bar,baz}
     \glossariesextrasetup{indexcrossrefs=true}
     \GlsXtrSetField{foo}{see}{bar,baz}

but with autoseeindex=false, this is equivalent to

     \newglossaryentry{foo}{name={foo},description={}}
     \GlsXtrSetField{foo}{see}{bar,baz}

Note that indexcrossrefs isn’t automatically implemented by the presence of the see key when autoseeindex is false.

It’s therefore possible to remove the cross-references from the location lists and set their position within the glossary style.

Another method of preventing the automatic indexing is to define the entries before the external indexing files have been opened with \makeglossaries. Since the appropriate file isn’t open, the information can’t be written to it. This will need the package option seenoindex=ignore (provided by glossaries) to prevent an error occurring.

record

(New to v1.08.) This is a ⟨key⟩=⟨value⟩ option provided for the benefit of bib2gls (see §10 bib2gls: Managing Reference Databases). If you want to use bib2gls, the recommended setting is record=nameref if you have hyperlinks and record=only if you don’t have hyperlinks.

The option may only be set in the preamble and can’t be used after \GlsXtrLoadResources. If the value is missing record=only is assumed. Permitted values:

off

This is the default setting. The indexing is performed as normal using either \makeglossaries or \makenoidxglossaries. This setting implements undefaction=error.

only

The indexing (recording) is performed by bib2gls (see §10 bib2gls: Managing Reference Databases). Neither \makeglossaries nor \makenoidxglossaries is permitted. This setting implements undefaction=warn and automatically loads the supplementary glossaries-extra-bib2gls package. (There should be no need to explicitly load glossaries-extra-bib2gls.)

The glossaries should be displayed using \printunsrtglossary (or \printunsrtglossaries).

The document build process is (assuming the file is called myDoc.tex):

         pdflatex myDoc
         bib2gls myDoc
         pdflatex myDoc

If you want letter groups you will need the --group or -g switch when invoking bib2gls:

         pdflatex myDoc
         bib2gls -g myDoc
         pdflatex myDoc

Note that record=only will prevent the see from automatically implementing \glssee. (bib2gls deals with the see field.) You may explicitly use \glssee in the document, but bib2gls will ignore the cross-reference if the see field was already set for that entry.

The record=only option will automatically set the glossaries package’s sort=none option if available. (That option value was only introduced to glossaries v4.30.)

nameref

(New to v1.37 and requires bib2gls v1.8+.) This option is like record=only but additionally records the current label information given by \@currentlabel and \@currentHref, and provides a more reliable way of saving the \theH⟨counter⟩ for the given location. This option requires hyperref otherwise it will fall back on the usual location records. Remember that \@currentHref is always globally updated whenever \refstepcounter is used, but \@currentlabel isn’t. This can cause some undesired side-effects with some settings. Remember also that the indexcounter option increments the associated counter every time an entry is indexed, which affects this option. See §10.3.2 Nameref Record for further details.

This option is best used with counter=chapter or counter=section if you want the title included in the location list. If the indexing counter is the default page, only the location number is shown. Similarly for counter=equation (or equations=true).

alsoindex

Deprecated synonym of hybrid.

hybrid

This is a hybrid setting that uses bib2gls to fetch entry information from .bib files, but uses makeindex or xindy to create the glossary files (which are input with \printglossary). Note that this requires a slower and more complicated build process (see below).

This hybrid approach is provided for the rare instances where an existing xindy rule or module is too complicated to convert to a bib2gls rule but the entries need to be fetched from a bib file. There’s no benefit in using this option with makeindex.

Since it’s redundant to make bib2gls also sort, use sort=none in \GlsXtrLoadResources for a faster build.

This option must be used with \makeglossaries but not with its optional argument. This option should not be used with \makenoidxglossaries. You may need to change the transcript file used by bib2gls to avoid a clash with the transcript file used by makeindex or xindy. (This can be done with bib2gls’s --log-file or -t option.)

Each glossary should be displayed using \printglossary (or \printglossaries for all of them). This option is expected to be used with bib2gls’s sort=none setting and so glossaries-extra-bib2gls is not automatically loaded.

The document build process is (assuming the file is called myDoc.tex):

         pdflatex myDoc
         bib2gls myDoc
         pdflatex myDoc
         makeglossaries myDoc
         pdflatex myDoc

Note that, in this case, it’s redundant to call bib2gls with the --group or -g switch as makeindex/xindy will insert the group heading information into the corresponding glossary file. (If you want bib2gls to form the letter groups then this hybrid method is inappropriate.)

With the recording on (record=only, record=nameref or record=hybrid), any of the commands that would typically index the entry (such as \gls, \glstext or \glsadd) will add a \glsxtr@record entry to the .aux file. bib2gls can then read these lines to find out which entries have been used. (Remember that commands like \glsentryname don’t index, so any use of these commands won’t add a corresponding \glsxtr@record entry to the .aux file.) See §10 bib2gls: Managing Reference Databases for further details.

The hybrid method additionally performs the standard indexing action that’s required for makeindex or xindy to work.

equations

(New to v1.37.) This option will cause the default location counter to automatically switch to equation when inside a numbered equation environment, such as equation or align. The counter can be explicitly overridden with counter in the optional arguments of commands like \glslink or \gls as usual.

floats

(New to v1.37.) This option will cause the default location counter to automatically switch to the corresponding counter when inside a floating environment, such as figure or table. The counter can be explicitly overridden with counter in the optional arguments of commands like \glslink or \gls as usual. Remember that within floats it’s the \caption command that actually uses \refstepcounter, so indexing before the caption will result in the wrong reference. The commands for use in captions and sections, such as \glsfmttext and \glsfmtshort, don’t index. (See §5 Entries in Sectioning Titles, Headers, Captions and Contents). You may want to consider using \glsadd after the caption (not before). For example:

     \begin{figure}[htbp]
       \centering
       \includegraphics{example-image}
       \caption{Sample \glsfmttext{foobar} figure}
       \glsadd{foobar}
     \end{figure}

indexcounter

(New to v1.29.) This option (which doesn’t take a value) is primarily intended for use with bib2gls (v1.4+) and hyperref. It can be used with makeindex or xindy but it will interfere with the number list collation, so you won’t have ranges and you’ll have duplicate page numbers present (but each page number will link to the relevant part of the page where the indexing occurred). This option automatically implements counter=wrglossary.

This option works by incrementing wrglossary and adding \label. This can cause a problem if the indexing occurs in an equation environment as amsmath forbids multiple occurrences of \label (resulting in the “Multiple \label’s” error). It’s best to change the counter to page or equation when in maths mode with this option. For example:

     \renewcommand{\glslinkpresetkeys}{%
      \ifmmode \setkeys{glslink}{counter=equation}\fi}
     \renewcommand{\glsaddpresetkeys}{%
      \ifmmode \setkeys{glossadd}{counter=equation}\fi}

By default (with hyperref), the page numbers in number lists link back to the top of the relevant page (provided the format uses \glshypernumber). The indexcounter option is designed to link back to the place within the page where the indexing occurred. It does this by creating a new counter (called wrglossary) that’s incremented with \refstepcounter every time an entry is indexed (but not via cross-referencing commands, such as \glssee). A \label is placed immediately after the increment command allowing the back-referenced to be obtained with \pageref. The location, as seen by the indexing application, is the value of the wrglossary counter but this value is substituted with the page reference when number list is typeset. Since the counter is used by all entries and is incremented every time any indexing occurs, neither makeindex nor xindy can correctly collate the lists. For example, if the first term to be referenced is indexed three times on page 5 without any intervening terms then the actual locations obtained from wrglossary will be 1, 2 and 3, which xindy and makeindex will try to form into the range 1–3, but they should actually all simply appear as page 5, whereas it can actually end up with 5–5. Conversely, a range may not be formed where it would naturally occur if just the page counter was used.

Since bib2gls is designed specifically to work with glossaries-extra, bib2gls (v1.4+) will check for wrglossary locations. If the default --merge-wrglossary-records is on, then any records for the same page for a given entry will be merged. In the above example with three references on page 5, only a single record for page 5 for that entry will be added to the number list and it will link back to the first instance on that page. Whereas if you use the --no-merge-wrglossary-records switch, the number list will contain three instance of page 5, with each linking to the corresponding place on that page. In both cases, consecutive pages can form ranges, but it may look strange in the second case.

See the bib2gls documentation for the save-index-counter resource option for more details.

docdef

This option governs the use of \newglossaryentry. It was originally a boolean option, but as from version 1.06, it can now take one of the following values (if the value is omitted, true is assumed):

• docdef=false: \newglossaryentry is not permitted in the document environment (default).
• docdef=true: \newglossaryentry behaves as it does in the base glossaries package. That is, where its use is permitted in the document environment, it uses the .glsdefs temporary file to store the entry definitions so that on the next LaTeX run the entries are defined at the beginning of the document environment. This allows the entry information to be referenced in the glossary, even if the glossary occurs before \newglossaryentry. (For example, when the glossary is displayed in the front matter.) This method of saving the definitions for the next LaTeX run has drawbacks that are detailed in the glossaries user manual.

  Remember that if \newglossaryentry wouldn’t be allowed in the document environment with the base glossaries package, then it still won’t be allowed with docdefs=true. If your glossaries occur at the end of the document, consider using docdef=restricted instead.

• docdef=restricted: (new to version 1.06) \newglossaryentry is permitted in the document environment without using the .glsdefs file. This means that all entries must be defined before the glossary is displayed, but it avoids the complications associated with saving the entry details in a temporary file. You will still need to take care about any changes made to characters that are required by the ⟨key⟩=⟨value⟩ mechanism (that is, the comma and equal sign) and any makeindex or xindy character that occurs in the sort key or label. If any of those characters are made active in the document, then it can cause problems with the entry definition. This option will allow \newglossaryentry to be used in the document with \makenoidxglossaries, but note that \longnewglossaryentry remains a preamble-only command.

  With this option, if an entry appears in the glossary before it has been defined, an error will occur (or a warning if the undefaction=warn option is used.) If you edit your document and either remove an entry or change its label, you may need to delete the document’s temporary files (such as the .aux and .gls files).

• docdef=atom: (new to version 1.34) This option behaves like docdef=restricted but creates the .glsdefs file for atom’s autocomplete support. This file isn’t input by glossaries-extra and so associated problems with the use of this file are avoided, but it allows the autocomplete support to find the labels in the file. As with docdef=restricted, entries may be defined in the preamble or anywhere in the document, but they may only be referenced after they have been defined. Entries must be defined before the associated glossary is displayed.

  If you need a list of all entry labels for the use of an editor or helper script you may also want to consider the package options writeglslabels and writeglslabelnames provided by the base glossaries package. Note that with these options and docdef=atom, only the entry labels visible to LaTeX can be saved. So if you are using bib2gls you will only get the labels of the entries that are selected by bib2gls.

The glossaries package allows \newglossaryentry within the document environment (when used with makeindex or xindy) but the user manual warns against this usage. By default the glossaries-extra package prohibits this, only allowing definitions within the preamble. If you are really determined to define entries in the document environment, despite all the associated drawbacks, you can restore this with docdef=true. Note that this doesn’t change the prohibitions that the glossaries package has in certain circumstances (for example, when using “option 1”). See the glossaries user manual for further details. A better option if document definitions are required is docdef=restricted. Only use docdef=true if document definitions are necessary and one or more of the glossaries occurs in the front matter.

This option affects commands that internally use \newglossaryentry, such as \newabbreviation, but not the “on-the-fly” commands described in §12 On-the-Fly Document Definitions.

nomissingglstext

This is a boolean option. If true, this will suppress the warning written to the transcript and the warning text that will appear in the document if the external glossary files haven’t been generated due to an incomplete document build. However, it’s probably simpler just to fix whatever has caused the failure to build the external file or files.

abbreviations

This option has no value and can’t be cancelled. If used, it will automatically create a new glossary with the label abbreviations and redefines \glsxtrabbrvtype to this label. (The file extensions are glg-abr, gls-abr and glo-abr.) In addition, this option defines a shortcut command

---

\printabbreviations[⟨options⟩]

---

which is equivalent to

\printglossary[type=\glsxtrabbrvtype,⟨options⟩]

If glossaries-extra-bib2gls is also loaded then this option will additionally provide:

---

\printunsrtabbreviations[⟨options⟩]

---

which uses \printunsrtglossary.

The title of the new glossary is given by

---

\abbreviationsname

---

If this command is already defined, it’s left unchanged. Otherwise it’s defined to “Abbreviations” if babel hasn’t been loaded or \acronymname if babel has been loaded. However, if you’re using babel it’s likely you will need to change this. (See §15 Multi-Lingual Support for further details.)

If you don’t use the abbreviations package option, the \abbreviationsname command won’t be defined (unless it’s defined by an included language file).

If the abbreviations option is used and the acronym option provided by the glossaries package hasn’t been used, then \acronymtype will be set to \glsxtrabbrvtype so that acronyms defined with \newacronym can be added to the list of abbreviations. If you want acronyms in the main glossary and other abbreviations in the abbreviations glossary then you will need to redefine \acronymtype to main:

     \renewcommand*{\acronymtype}{main}

Note that there are no analogous options to the glossaries package’s acronymlists option (or associated commands) as the abbreviation mechanism is handled differently with glossaries-extra.

symbols

This is passed to glossaries but will additionally define

---

\glsxtrnewsymbol[⟨options⟩]{⟨label⟩}{⟨symbol⟩}

---

which is equivalent to

\newglossaryentry{⟨label⟩}{name={⟨symbol⟩},
 sort={⟨label⟩},type=symbols,category=symbol,⟨options⟩}

Note that the sort key is set to the ⟨label⟩ not the ⟨symbol⟩ as the symbol will likely contain commands.

If glossaries-extra-bib2gls is also loaded then this option will additionally provide:

---

\printunsrtsymbols[⟨options⟩]

---

which uses \printunsrtglossary.

numbers

This is passed to glossaries but will additionally define

---

\glsxtrnewnumber[⟨options⟩]{⟨number⟩}

---

which is equivalent to

\newglossaryentry{⟨label⟩}{name={⟨number⟩},
 sort={⟨label⟩},type=numbers,category=number,⟨options⟩}

If glossaries-extra-bib2gls is also loaded then this option will additionally provide:

---

\printunsrtnumbers[⟨options⟩]

---

which uses \printunsrtglossary.

acronyms (or acronym)

This is passed to glossaries but if glossaries-extra-bib2gls is also loaded then this option will additionally provide:

---

\printunsrtacronyms[⟨options⟩]

---

which uses \printunsrtglossary.

This option defines a new glossary with the label acronym not acronyms. You may find it easier to reference it with the command \acronymtype.

index

This is passed to glossaries but if glossaries-extra-bib2gls is also loaded then this option will additionally provide:

---

\printunsrtindex[⟨options⟩]

---

which uses \printunsrtglossary.

shortcuts

Unlike the glossaries package option of the same name, this option isn’t boolean but has multiple values:

• shortcuts=acronyms (or shortcuts=acro): set the shortcuts provided by the glossaries package for acronyms (such as \ac). Note that the short and long forms don’t use \glsxtrshort and \glsxtrlong but use the original \acrshort and \acrlong, which don’t recognise multiple abbreviation styles. The better option with glossaries-extra is shortcuts=ac.
• shortcuts=ac: set the shortcuts provided by the glossaries package for acronyms (such as \ac) but uses the glossaries-extra interface (such as \glsxtrshort rather than \acrshort). In this case \ac is defined as \cgls rather than \gls.
• shortcuts=abbreviations (or shortcuts=abbr): set the abbreviation shortcuts provided by glossaries-extra. (See §4.3 Shortcut Commands.) These settings don’t switch on the acronym shortcuts provided by the glossaries package.
• shortcuts=other: set the “other” shortcut commands, but not the shortcut commands for abbreviations or the acronym shortcuts provided by glossaries. The “other” shortcuts are:
  • \newentry equivalent to \newglossaryentry
  • \newsym equivalent to \glsxtrnewsymbol (see the symbols option).
  • \newnum equivalent to \glsxtrnewnumber (see the numbers option).
• shortcuts=all (or shortcuts=true): implements shortcuts=ac, shortcuts=abbreviations and shortcuts=other.
• shortcuts=none (or shortcuts=false): don’t define any of the shortcut commands (default).

Note that multiple invocations of the shortcuts option within the same option list will override each other.

After the glossaries-extra package has been loaded, you can set available options using

---

---

The abbreviations and docdef options may only be used in the preamble. Additionally, docdef can’t be used after \makenoidxglossaries.

2. Modifications to Existing Commands and Styles

Top

2.1 Defining Entries

The glossaries package provides \nopostdesc which may be used in the description to suppress the post-description hook. The glossaries-extra package provides another command

---

---

which has a similar function but only suppresses the post-description punctuation. It doesn’t suppress the use of \glsxtrpostdescription which allows the use of category-dependent post-description hooks. (Note that the punctuation, which is in the original base hook \glspostdescription, comes after the extended post-description hook \glsxtrpostdescription not before.) The post-description hook can counter-act the effect of \glsxtrnopostpunc using

---

---

These commands have no effect outside of the glossary (except with standalone entries that use \glsxtractivatenopost and \glspostdescription, see §11.4 Standalone Entry Items).

The commands used by glossaries to automatically produce an error if an entry is undefined (such as \glsdoifexists) are changed to take the undefaction option into account.

The \newglossaryentry command has three new keys:

• category, which sets the category label for the given entry. By default this is general. See §7 Categories for further information about categories.
• alias, which allows an entry to be alias to another entry. See §11.5 Entry Aliases for further details.
• seealso, which performs much like see, but allows for separate “see” and “see also” treatment. See §2.3 Cross-References (“see” and “see also”) for further details.

This apply to all entry defining commands (such as \newabbreviation).

The test file example-glossaries-xr.tex contains dummy entries with a mixture of see, alias and seealso keys for use with minimal working examples. There are also example-glossaries-*.bib files that correspond to each example-glossaries-*.tex file for testing bib2gls.

The \longnewglossaryentry command now has a starred version (as from v1.12) that doesn’t automatically insert

\leavevmode\unskip\nopostdesc

at the end of the description field.

---

---

The descriptionplural key is left unset unless explicitly set in ⟨options⟩.

The unstarred version no longer hard-codes the above code (which removes trailing space and suppresses the post-description hook) but instead uses:

---

---

This can be redefined to allow the post-description hook to work but retain the \unskip part if required. For example:

\renewcommand*{\glsxtrpostlongdescription}{\leavevmode\unskip}

This will discarded unwanted trailing space at the end of the description but won’t suppress the post-description hook.

The unstarred version also alters the base glossaries package’s treatment of the descriptionplural key. Since a plural description doesn’t make much sense for multi-paragraph descriptions, the default behaviour with glossaries-extra’s \longnewglossaryentry is to simply leave the plural description unset unless explicitly set using the descriptionplural key. The glossaries.sty version of this command sets the description’s plural form to the same as the singular.^2.1

Note that this modified unstarred version doesn’t append \glsxtrpostlongdescription to the description’s plural form.

The \newterm command (defined through the index package option) is modified so that the category defaults to index. The \newacronym command is modified to use the new abbreviation interface provided by glossaries-extra. (See §4 Abbreviations.)

The \makeglossaries command now has an optional argument.

---

---

If ⟨list⟩ is empty, \makeglossaries behaves as per its original definition in the glossaries package, otherwise ⟨list⟩ can be a comma-separated list of glossaries that need processing with an external indexing application.

This command is not permitted with the record=only package option. Without the optional argument, it’s required with record=hybrid. With the optional argument, it’s only permitted with the default record=off.

It should then be possible to use \printglossary for those glossaries listed in ⟨list⟩ and \printnoidxglossary for the other glossaries. (See the accompanying file sample-mixedsort.tex for an example.)

You will need at least version 2.20 of makeglossaries or at least version 1.3 of the Lua alternative makeglossaries-lite (both distributed with glossaries v4.27) to allow for this use of \makeglossaries[⟨list⟩]. Alternatively, use the automake option.

Top

2.2 Entry Indexing

As from version 1.31, there is a new command like \glsadd where the mandatory argument is a comma-separated list of labels:

---

---

This simply iterates over ⟨list⟩ and does \glsadd[⟨options⟩]{⟨label⟩} for each entry in the list.

As from version 1.37, you can make commands like \gls or \glslink automatically use \glsadd with specific options when a given format is used (identified with format={⟨format⟩} in the optional argument of the corresponding \gls, \glslink etc).

---

---

The optional argument ⟨label⟩ defaults to \glslabel and indicates the label to use in \glsadd and so needs to be expandable. The ⟨format list⟩ is a comma-separated list of format values that will trigger the automated adding. The ⟨glsadd options⟩ are the options to pass to \glsadd with format={⟨format⟩} prepended to the list.

For example, with:

\GlsXtrAutoAddOnFormat{hyperbf}{counter=chapter}

then \gls[format=hyperbf]{sample} will be equivalent to

\glsadd[format=hyperbf,counter=chapter]{sample}\gls[format=hyperbf]{sample}

Note that the explicit range markers will prevent a match unless you include them in ⟨format list⟩ (in which case, be sure to add both the start and end formats).

Here’s another example:

\GlsXtrAutoAddOnFormat[dual.\glslabel]{hyperbf}{}

In this case \gls[format=hyperbf]{sample} will now be equivalent to:

\glsadd[format=hyperbf]{dual.sample}\gls[format=hyperbf]{sample}

The glossaries-extra package provides extra keys for commands like \gls and \glstext:

noindex

This is a boolean key. If true, this suppresses the indexing. (That is, it prevents \gls or whatever from adding a line to the external glossary file.) This is more useful than the indexonlyfirst package option provided by glossaries, as the first use might not be the most pertinent use. (If you want to apply indexonlyfirst to selected entries, rather than all of them, you can use the indexonlyfirst attribute, see §7 Categories for further details.) Note that the noindex key isn’t available for \glsadd (and \glsaddall) since the whole purpose of that command is to index an entry.

wrgloss

(New to v1.14.) This is may only take the values before or after. By default, commands that both index and display link text (such as \gls, \glstext and \glslink), perform the indexing before the link text as the indexing creates a whatsit that can cause problems if it occurs after the link text. However, it may be that in some cases (such as long phrases) you may actually want the indexing performed after the link text. In this case you can use wrgloss=after for specific instances. Note that this option doesn’t have an effect if the indexing has been suppressed through other settings (such as noindex).

The default value is set up using

---

\glsxtrinitwrgloss

---

which is defined as:

     \newcommand*{\glsxtrinitwrgloss}{%
      \glsifattribute{\glslabel}{wrgloss}{after}%
      {%
        \glsxtrinitwrglossbeforefalse
      }%
      {%
        \glsxtrinitwrglossbeforetrue
      }%
     }

This sets the conditional

---

\ifglsxtrinitwrgloss

---

which is used to determine where to perform the indexing.

This means you can set the wrgloss attribute to after to automatically use this as the default for entries with that category attribute. (Note that adding wrgloss to the default options in \GlsXtrSetDefaultGlsOpts will override \glsxtrinitwrgloss.)

hyperoutside

(New to v1.21.) This is a boolean key. The default is hyperoutside=true, which puts the hyperlink outside \glstextformat, so that commands like \gls will effectively do

\hyperlink{⟨target⟩}{\glstextformat{⟨link text⟩}}

This is the same behaviour as with the base glossaries package. With hyperoutside=false, \hyperlink is placed inside the argument of \glstextformat:

\glstextformat{\hyperlink{⟨target⟩}{⟨link text⟩}}

You can use the hyperoutside category attribute to set the default for a given category. This can be combined with the use of the textformat attribute to counteract any interference caused by \hyperlink. For example:

     \glssetcategoryattribute{mathrelation}{hyperoutside}{false}

will set hyperoutside=false for all entries that are assigned to the category mathrelation and

     \glssetcategoryattribute{mathrelation}{textformat}{mathrel}

will use \mathrel instead of \glstextformat resulting in:

\mathrel{\hyperlink{⟨target⟩}{⟨link text⟩}}

for entries with the category key set to mathrelation.

textformat

This key must have a control sequence name as its value. The command formed from this name must exist and must take one argument. (Use relax for default behaviour.) If set, this overrides the textformat attribute and \glstextformat. See the soul example in §2.6 First Use Flag.

prefix

Locally redefines \glolinkprefix to the given value. It should match the prefix for the desired glossary.

thevalue

Explicitly set the location to this value (see below).

theHvalue

Set the corresponding hyperlink location (see below).

You can set the default options used by \glslink, \gls etc with:

---

---

For example, if you mostly don’t want to index entries then you can do:

\GlsXtrSetDefaultGlsOpts{noindex}

and then use, for example, \gls[noindex=false]{sample} when you actually want the location added to the number list. These defaults may be overridden by other settings (such as category attributes) in addition to any settings passed in the option argument of commands like \glslink and \gls.

Note that if you don’t want any indexing, just omit \makeglossaries and \printglossaries (or analogous commands). If you want to adjust the default for wrgloss, it’s better to do this by redefining \glsxtrinitwrgloss instead.

If you want to change the default value of format, you can instead use:

---

---

This has the advantage of also working for \glsadd. For example, if you want all locations in the back matter to appear in italic (unless explicitly overridden):

\backmatter
\GlsXtrSetDefaultNumberFormat{hyperit}

Commands like \gls have star (*) and plus (+) modifiers as a short cut for hyper=false and hyper=true. The glossaries-extra package provides a way to add a third modifier, if required, using

---

---

where ⟨char⟩ is the character used as the modifier and ⟨options⟩ is the default set of options (which may be overridden). Note that ⟨char⟩ must be a single character (not a UTF-8 character, unless you are using XƎLATEX or LuaLATEX).

Example:

\GlsXtrSetAltModifier{!}{noindex}

This means that \gls!{sample} will be equivalent to \gls[noindex]{sample}. It’s not possible to mix modifiers. For example, if you want to do

\gls[noindex,hyper=false]{sample}

you can use \gls*[noindex]{sample} or \gls![hyper=false]{sample} but you can’t combine the * and ! modifiers.

There is a new hook that’s used each time indexing information is written to the external glossary files:

---

---

where ⟨label⟩ is the entry’s label. This does nothing by default but may be redefined. (See, for example, the accompanying sample file sample-indexhook.tex, which uses this hook to determine which entries haven’t been indexed.)

There’s also a new hook (from v1.26) that’s used immediately before the options are set by the \gls-like and \glstext-like commands:

---

---

(The base package provides \glslinkpostsetkeys that’s used immediately after the options are set.)

As from version 1.30 there are also similar hooks for \glsadd:

---

---

and

---

---

For example, to default to using the equation counter in maths mode:

\renewcommand{\glslinkpresetkeys}{%
 \ifmmode \setkeys{glslink}{counter=equation}\fi}
\renewcommand{\glsaddpresetkeys}{%
 \ifmmode \setkeys{glossadd}{counter=equation}\fi}

In this case, the counter can be overridden with an explicit use of counter in the optional argument of \gls or \glsadd. (As from version 1.37, a simpler method is to just use the equations package option.)

Alternatively, to enforce this (overriding the option argument):

\renewcommand{\glslinkpostsetkeys}{%
 \ifmmode \setkeys{glslink}{counter=equation}\fi}
\renewcommand{\glsaddpostsetkeys}{%
 \ifmmode \setkeys{glossadd}{counter=equation}\fi}

As from version 1.14, there are two new keys for \glsadd: thevalue and theHvalue. These keys are designed for manually adding explicit locations rather than obtaining the value from the associated counter. As from version 1.19, these two keys are also available for commands like \gls and \glslink. The thevalue keys is intended primarily for adding locations in supplementary material that can’t be obtained from a counter.

The principle key thevalue is for the location value. The other key theHvalue can be used to extract a prefix value for the first argument of commands like \glsnoidxdisplayloc. Its value must be in the format ⟨prefix⟩⟨location⟩. In general, there’s little need for this key as the prefix is typically associated with a counter that can be used to form hypertargets.

For example, makeindex will only accept locations in the form [⟨num⟩⟨sep⟩]*⟨num⟩ where ⟨num⟩ is an arabic number (0, 1, …), roman numeral (i, ii, … or I, II, …) or a character from a, …, z or A, …, Z, and [⟨num⟩⟨sep⟩]* indicates zero or more instances of a number followed by the recognised separator character (set with \glsSetCompositor). This means that makeindex won’t accept, for example,

\glsadd[thevalue={Supplementary Material}]{sample}

This location value will be accepted by bib2gls, since it will allow any location and will only try forming ranges if the location matches any of its numerical patterns. In the case of xindy, you’ll need to add a rule that can match the value. If you’re using hyperref, you may need to use the format key to prevent a hyperlink if one can’t naturally be formed from the prefix, counter and location value.

For example, suppose the file suppl.tex contains:

\documentclass{article}
\usepackage{glossaries-extra}
\newglossaryentry{sample}{name={sample},description={an example}}
\renewcommand{\thepage}{S.\arabic{page}}
\begin{document}
First page.
\newpage
\gls{sample}.
\end{document}

This has an entry on page S.2. Suppose another document wants to include this location in the glossary. Then this can be done by setting thevalue to S.2. For example:

\documentclass{article}
\usepackage{glossaries-extra}
\makeglossaries
\newglossaryentry{sample}{name={sample},description={an example}}
\begin{document}
Some \gls{sample} text.
\printglossaries
\glsadd[thevalue={S.2}]{sample}
\end{document}

This location value will be accepted by makeindex as it’s in the form ⟨num⟩⟨sep⟩⟨num⟩.

If you want hyperlinks, things are more complicated. First you need to set the externallocation attribute to the location of the PDF file. For example:

\glssetcategoryattribute{supplemental}{externallocation}{suppl.pdf}
\newglossaryentry{sample}{category=supplemental,
 name={sample},description={an example}}

Next you need to add glsxtrsupphypernumber as the format:

\glsadd[thevalue={S.2},format=glsxtrsupphypernumber]{sample}

Both documents will need to use the hyperref package. Remember that the counter used for the location also needs to match. If \theH⟨counter⟩ is defined in the other document and doesn’t match in the referencing document, then you need to use theHvalue to set the appropriate value. See the accompanying sample files sample-suppl-hyp.tex and sample-suppl-main-hyp.tex for an example that uses hyperref.

For example, if both sample-suppl-hyp.pdf and sample-suppl-main-hyp.pdf are in the same directory, then viewing sample-suppl-main-hyp.pdf in Evince will take you to the correct location in the linked document (when you click on the S.2 external link), but Okular will take you to the top of the first page of the linked document.

This method can only be used where there is one external source for the designated category (identified by the externallocation attribute). For multiple sources, you need to use bib2gls version 1.7+, which is the better method in general as it can automatically fetch the relevant locations from the .aux files of the designated external documents without the need to explicitly use \glsadd.

Top

2.3 Cross-References (“see” and “see also”)

The value of the see key is now saved as a field. This isn’t the case with glossaries, where the see value is simply used to directly write a line to the corresponding glossary file and is then discarded. This is why the see key can’t be used before \makeglossaries (since the file hasn’t been opened yet).

This modification allows glossaries-extra to provide

---

---

which is used at the end of the document to automatically add any unused cross-references unless the package option indexcrossrefs was set to false.

Note that even though the see key will now work for entries defined in the document environment, it’s still best to define entries in the preamble, and the see key still can’t perform any indexing before the file has been opened by \makeglossaries. Note that glossaries v4.24 introduced the seenoindex package option, which can be used to suppress the error when the see key is used before \makeglossaries, so seenoindex=ignore will allow the see value to be stored even though it may not be possible to index it at that point.

As from version 1.06, you can display the cross-referenced information for a given entry using

---

---

This internally uses

---

---

where ⟨tag⟩ and ⟨xr list⟩ are obtained from the value of the entry’s see field (if non-empty). By default, this just does \glsseeformat[⟨tag⟩]{⟨xr list⟩}{}, which is how the cross-reference is displayed in the number list. Note that \glsxtrusesee does nothing if the see field hasn’t been set for the entry given by ⟨label⟩.

As with the base glossaries package, \glsseeformat is defined to do \emph{⟨tag⟩} \glsseelist{⟨xr list⟩}. The third argument is always ignored and is present for makeindex, which always requires a final argument to encapsulate the associated location. The command

---

---

used to iterate over the list of cross-reference labels is also unchanged from the base glossaries package, with each item in the list formatted according to:

---

---

This is defined by the glossaries package to:

So the actual formatting for each cross-referenced entry is performed by \glsseeitemformat, which is redefined by glossaries-extra, as described in §2 Modifications to Existing Commands and Styles. This now displays the value of the text field for abbreviations and the value of the name field otherwise. There’s no indication of the entry hierarchy, which could confuse the reader. Therefore, as from glossaries-extra v1.37, there are some new commands that include the hierarchical information. You may prefer to redefine \glsseeitemformat to use one of this if you have sub-entries. For example:

\renewcommand*{\glsseeitemformat}[1]{\glsxtrhiername{#1}}

The glossaries package provides

---

---

to format items in a cross-reference list (identified with the see key or \glssee). This was originally defined to use \glsentryname{⟨label⟩} since it makes more sense for the cross-reference to match the way the term appears in the glossary. Unfortunately this caused a problem when the name field was sanitized, which used to be the default setting, so glossaries v3.0 changed the default definition of this command to use \glsentrytext instead. Since the name and text field are quite often the same, this change usually doesn’t have a noticeable effect. However, now that the name field is no longer sanitized (following the redesign of glossaries v4.0) it makes more sense to restore this command to its original behaviour, but to take account of abbreviations glossaries-extra redefines this as:

\renewcommand*{\glsseeitemformat}[1]{%
 \ifglshasshort{\glslabel}{\glsfmttext{#1}}{\glsfmtname{#1}}%
}

(Note that as from glossaries-extra version 1.42, this now uses \glsfmttext and \glsfmtname instead of just referencing the text and name fields. This helps to ensure that any formatting is correctly applied.)

If you want to restore the glossaries v3.0+ definition just do:

\renewcommand*{\glsseeitemformat}[1]{\glsentrytext{#1}}

The glossaries-extra package provides \glsxtrhiername and its case-changing variants that may be used within the definition of \glsseeitemformat if required. These display the hierarchy for sub-entries rather than just the name, which may be more helpful in cross-references.

---

---

performs a recursive action:

1. If the entry given by ⟨label⟩ has a parent, then \glsxtrhiername{⟨parent label⟩} is done followed by \glsxtrhiernamesep then:
2. If the entry given by ⟨label⟩ is an abbreviation (that is, it has the short field set) then the short form is displayed (using \glsfmtshort) otherwise the name is displayed (using \glsfmtname).

The first step above is skipped if the entry doesn’t have a parent. Each level is separated by:

---

---

which defaults to “ ⊳ ”. This can be redefined as appropriate.

There are some case-changing variants:

---

---

The top-level has the first letter changed to upper case (either \Glsfmtshort or \Glsfmtname). There’s no case-change for sub-entries.

---

---

All levels have the first letter changed to upper case (either \Glsfmtshort or \Glsfmtname).

---

---

The top-level is converted to upper case (either \GLSfmtshort or \GLSfmtname). There’s no case-change for sub-entries.

---

---

All levels are converted to upper case (either \GLSfmtshort or \GLSfmtname).

Suppose you want to suppress the number list using nonumberlist. This will automatically prevent the cross-references from being displayed. The seeautonumberlist package option will automatically enable the number list for entries that have the see key set, but this will also show the rest of the number list.

Another approach in this situation is to use the post description hook with \glsxtrusesee to append the cross-reference after the description. For example:

\renewcommand*{\glsxtrpostdescgeneral}{%
 \ifglshasfield{see}{\glscurrententrylabel}
 {, \glsxtrusesee{\glscurrententrylabel}}%
 {}%
}

Now the cross-references can appear even though the number list has been suppressed.

As from v1.16, there’s a separate seealso key. Unlike see, this doesn’t have an optional part for the textual tag. The syntax seealso={⟨xr-labels⟩} works in much the same way as using see=[\seealsoname]{⟨xr-labels⟩} but the information is stored in a separate field. If you need a different tag, use the see key instead (or redefine \seealsoname or \glsxtruseseealsoformat, described below).

You can display the formatted list of cross-references stored in the seealso key using:

---

---

This works in much the same way as \glsxtrusesee but it internally uses

---

---

For example:

\renewcommand*{\glsxtrpostdescgeneral}{%
 \ifglshasfield{see}{\glscurrententrylabel}
 {, \glsxtrusesee{\glscurrententrylabel}}%
 {}%
 \ifglshasfield{seealso}{\glscurrententrylabel}
 { (\glsxtruseseealso{\glscurrententrylabel})}%
 {}%
}

The alias key only contains a single label not a list, but if you want to ensure consistent formatting with \glsxtrusesee and \glsxtruseseealso you can use (from v1.42):

---

---

The actual unformatted comma-separated list ⟨xr-list⟩ stored in the seealso field can be accessed with:

---

---

This will just expand to the ⟨xr-labels⟩ provided in the value of the seealso key. There’s no corresponding command to access the see field. If you really need to access it, you can use commands like \glsxtrfielduse, but remember that it may start with [⟨tag⟩], so it can’t be automatically treated as a simple comma-separated list.

As mentioned above, the base glossaries package provides \glsseelist, which requires a comma-separated list of labels as the argument. The argument isn’t fully expanded, so it’s not suitable to use, for example, \glsxtrseealsolabels{⟨label⟩} as the argument. For convenience, glossaries-extra provides

---

---

which fully expands its argument and passes it to \glsseelist.

As from v1.47, glossaries-extra redefines \glsseelist to make it more flexible. The base package provides \glsseeitem{⟨label⟩} which is used to format each element in the list and \glsseelastsep for the separator between the final two items (\glsseesep is used between the other items). The modifications provide two additional commands:

---

---

This defaults to \glsseeitem{⟨label⟩} and is used to format the first label in the list.

---

---

This defaults to \glsseelastsep and is used if the list consists of three or more labels.

The seealso key implements the automatic indexing using

---

---

which just does

unless the xindy option is used with glossaries v4.30+, in which case a distinct seealso cross-reference class is used instead.

The command that produces this “see also” text is

---

---

If \alsoname (provided by language packages) is defined then \seealsoname will simply be defined to \alsoname otherwise it will be defined to see also.

The language-sensitive \alsoname is used by general indexing packages, such as makeidx, so if you redefine \alsoname the change will apply to normal indexes as well as glossaries. If you only want to change the text produced by the seealso key without affecting general indexing (with \index) then redefine \seealsoname instead.

Top

2.4 Entry Display Style Modifications

Recall from the glossaries package that commands such as \gls display text at that point in the document (optionally with a hyperlink to the relevant line in the glossary). This text is referred to as the “link-text” regardless of whether or not it actually has a hyperlink. The actual text and the way it’s displayed depends on the command used (such as \gls) and the entry format.

The default entry format (\glsentryfmt) used in the link-text by commands like \gls, \glsxtrfull, \glsxtrshort and \glsxtrlong (but not commands like \glslink, \glsfirst and \glstext) is changed by glossaries-extra to test for regular entries, which are determined as follows:

• If an entry is assigned to a category that has the regular attribute set (see §7 Categories), the entry is considered a regular entry, even if it has a value for the short key. In this case \glsentryfmt uses \glsgenentryfmt (provided by glossaries), which uses the first (or firstplural) value on first use and the text (or plural) value on subsequent use.
• An entry that doesn’t have a value for the short key is assumed to be a regular entry, even if the regular attribute isn’t set to “true” (since it can’t be an abbreviation without the short form). In this case \glsentryfmt uses \glsgenentryfmt.
• If an entry does has a value for the short key and hasn’t been marked as a regular entry through the regular attribute, it’s not considered a regular entry. In this case \glsentryfmt uses \glsxtrgenabbrvfmt (defined by glossaries-extra) which is governed by the abbreviation style (see §4.2 Abbreviation Styles).

This means that entries with a short form can be treated as regular entries rather than abbreviations if it’s more appropriate for the desired style.

As from version 1.04, \glsentryfmt now puts \glsgenentry in the argument of the new command

---

---

This just does its argument ⟨text⟩ by default. This means that if you want regular entries in a different font but don’t want that font to apply to abbreviations, then you can redefine \glsxtrregularfont. This is more precise than changing \glstextformat which is applied to all linking commands for all entries, unless overridden by the textformat attribute.

For example:

\renewcommand*{\glsxtrregularfont}[1]{\textsf{#1}}

You can access the label through \glslabel. For example, you can query the category:

\renewcommand*{\glsxtrregularfont}[1]{%
 \glsifcategory{\glslabel}{general}{\textsf{#1}}{#1}}

or query the category attribute, for example, provide a custom attribute called font:

\glssetcategoryattribute{general}{font}{sf}
\renewcommand*{\glsxtrregularfont}[1]{%
 \glsifattribute{\glslabel}{font}{sf}{\textsf{#1}}{#1}}

As from version 1.21, it’s simpler to just do, for example:

\glssetcategoryattribute{general}{textformat}{textsf}

without redefining \glsxtrregularfont.

As from version 1.30, there is also a command for abbreviations that encapsulates \glsxtrgenabbrvfmt:

---

---

This also just does its argument by default. Font changes made by abbreviation styles are within ⟨text⟩.

The \glspostlinkhook provided by the glossaries package to insert information after the link-text produced by commands like \gls and \glstext is redefined to

---

---

This command will discard a following full stop (period) if the discardperiod attribute is set to “true” for the current entry’s category. It will also do

---

---

if a full stop hasn’t be discarded and

---

---

if a full stop has been discarded.

It may be that you want to check some other setting (rather than a category attribute) to determine whether or not to discard a following full stop. In which case you can redefine:

---

---

You can access the field’s label using \glslabel. This command should do ⟨true⟩ if the post-link hook should check if a period follows and ⟨false⟩ otherwise. The default definition is simply:

\newcommand*{\glsxtrifcustomdiscardperiod}[2]{#2}

which means that no additional checks are performed. (Only the recognised category attributes will be checked.)

By default \glsxtrpostlink just does \glsxtrpostlink⟨category⟩ if it exists, where ⟨category⟩ is the category label for the current entry. (For example, for the general category, \glsxtrpostlinkgeneral if it has been defined.)

You can define the post-link hook command using \newcommand, for example:

\newcommand*{\glsxtrpostlinkgeneral}{%
 \glsxtrpostlinkAddDescOnFirstUse
}

or, as from v1.31, you can use

---

---

This is simply a shortcut for:

Note that it doesn’t check if the command has already been defined.

The sentence-ending hook is slightly more complicated. If the command \glsxtrpostlink⟨category⟩ is defined the hook will do that and then insert a full stop with the space factor adjusted to match the end of sentence. If \glsxtrpostlink⟨category⟩ hasn’t been defined, the space factor is adjusted to match the end of sentence. This means that if you have, for example, an entry that ends with a full stop, a redundant following full stop will be discarded and the space factor adjusted (in case the entry is in uppercase) unless the entry is followed by additional material, in which case the following full stop is no longer redundant and needs to be reinserted.

There are some convenient commands you might want to use when customizing the post-link-text category hooks:

---

---

This will add the description in parentheses on first use.

For example, suppose you want to append the description in parentheses on first use for entries in the symbol category:

\newcommand*{\glsxtrpostlinksymbol}{%
  \glsxtrpostlinkAddDescOnFirstUse
}

---

---

This will append the symbol (if defined) in parentheses on first use. (Does nothing if the symbol hasn’t been set.)

---

---

(New to v1.31.) On first use, this will append \space(⟨symbol⟩, ⟨description⟩) if the symbol is defined otherwise it just appends \space(⟨description⟩).

If you want to provide your own custom format be aware that you can’t use \ifglsused within the post-link-text hook as by this point the first use flag will have been unset. Instead you can use

---

---

This will do ⟨true⟩ if the last used entry was the first use for that entry, otherwise it will do ⟨false⟩. (Requires at least glossaries v4.19 to work properly.) This command is locally set by commands like \gls, so don’t rely on it outside of the post-link-text hook.

For example, if you want to place the description in a footnote after the link-text on first use for the general category:

\newcommand*{\glsxtrpostlinkgeneral}{%
  \glsxtrifwasfirstuse{\footnote{\glsentrydesc{\glslabel}}}{}%
}

The short-postfootnote abbreviation style uses the post-link-text hook to place the footnote after trailing punctuation characters.

Top

2.5 Entry Counting Modifications

If you are using bib2gls you may find it more convenient to use the record count commands described in §10 bib2gls: Managing Reference Databases instead.

The \glsenableentrycount command is modified to allow for the entrycount attribute. This means that you not only need to enable entry counting with \glsenableentrycount, but you also need to set the appropriate attribute (see §7 Categories).

For example, instead of just doing:

\glsenableentrycount

you now need to do:

\glsenableentrycount
\glssetcategoryattribute{abbreviation}{entrycount}{1}

This will enable the entry counting for entries in the abbreviation category, but any entries assigned to other categories will be unchanged.

Further information about entry counting, including the new per-unit feature, is described in §8.1 Entry Counting (First Use Flag).

Top

2.6 First Use Flag

The glossaries package provides

---

---

to determine whether or not an entry has been used. This requires the entry to have been defined. If the entry is undefined, then the underlying boolean variable doesn’t exist and so is neither true nor false. This command will produce an error with the default undefaction=error mode and a warning with undefaction=warn and, where appropriate, displays ?? in the document text to denote an undefined reference. With both modes, neither ⟨true⟩ nor ⟨false⟩ will be performed if ⟨label⟩ doesn’t exist.

The record option automatically implements undefaction=warn since no entries will be defined until they have been selected by bib2gls. In this case, you may prefer to use:

---

---

(new to v1.34) which does ⟨true⟩ if the entry given by ⟨label⟩ is undefined or if the entry is marked as unused. Remember that neither \ifglsused nor \GlsXtrIfUnusedOrUndefined should be used in the post-link hook as the first use flag will have already been unset before the hook is performed. (Instead, \glsxtrifwasfirstuse should be used, see §2.4 Entry Display Style Modifications.)

There are two new commands provided with version 1.31+:

---

---

and

---

---

These behave like \glslocalreset and \glslocalunset but the argument is a comma-separated list of labels.

The internal command used by \glsunset is modified first to allow for the changing in entry counting, described above, but also to allow for buffering pending unsets when commands like \gls are used in a context where changing a boolean variable can cause things to go wrong. One example of this is using \gls in one of the commands provided with the package. For example:

\ul{Some text about \gls{html}.}

This causes the confusing error:

Glossary entry `{html}' has not been defined.

The simplest workaround is to put \gls{html} inside the argument of \mbox. For example:

\ul{Some text about \mbox{\gls{html}}.}

This can work provided it’s not the first use of this entry. It if is, then unsetting the first use flag causes a problem and results in the error:

! Package soul Error: Reconstruction failed.

The glossaries-extra package provides a way of temporarily switching off \glsunset so that it just makes a note of the entry’s label but doesn’t actually perform the change:

---

---

The unstarred version doesn’t check for duplicates, so the internal list may end up with multiple occurrences of the same label. The starred version only adds a label to the internal list if it’s not already in it. Note that this buffering only applies to the global \glsunset and does not affect the local \glslocalunset.

Later you can restore \glsunset and unset all buffered labels using:

---

---

The starred form \GlsXtrStopUnsetBuffering* uses \glslocalunset instead. For example:

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage{soul}
\usepackage{glossaries-extra}
\newabbreviation{html}{HTML}{hypertext markup language}
\begin{document}
\GlsXtrStartUnsetBuffering
\ul{Some text about \mbox{\gls{html}}.}
\GlsXtrStopUnsetBuffering
Next use: \gls{html}.
\end{document}

Before you stop the unset buffering, you can iterate over the current buffer using

---

---

where ⟨cs⟩ is a control sequence that takes a single argument (which is the entry label). This is best used with the starred version \GlsXtrStartUnsetBuffering* to avoid duplicates.

You can discard the buffer and restore \glsunset to its normal behaviour with (new to v1.42):

---

---

There can still be a problem here as content within \mbox can’t break across a line so you may end up with an overfull line or excessive white space within the paragraph.

An alternative is to use \protect:

\GlsXtrStartUnsetBuffering
\ul{Some text about \protect\gls{html}.}
\GlsXtrStopUnsetBuffering

but the formatting (underlining in this example) won’t be applied. Another possibility is:

\usepackage[T1]{fontenc}
\usepackage{soul}
\usepackage{glossaries-extra}
\newabbreviation{html}{HTML}{hypertext markup language}
\newrobustcmd{\gul}[1]{%
  {%
   \def\glsxtrabbreviationfont##1{\GlsXtrExpandedFmt{\ul}{##1}}%
   \def\glsxtrregularfont##1{\GlsXtrExpandedFmt{\ul}{##1}}%
   #1%
  }%
}
\begin{document}
\ul{Some text about }\gls[textformat=gul]{html}.
Next use: \gls{html}.
\end{document}

This moves \gls outside of \ul and uses textformat to locally change the formatting command used by \gls (which is normally given by \glstextformat or the textformat attribute) to a custom command \gul, which locally changes the regular font and the default abbreviation font to use \ul. It then uses

---

---

which (protected) fully expands ⟨text⟩ before applying ⟨cs⟩, which must be a control sequence that takes a single argument. This allows \ul to move much further inside and increases its chances of working. It can still break if ⟨text⟩ expands to something that \ul can’t deal with. For example, if an abbreviation style is used that contains complex formatting or if the field value contains problematic content.

Top

2.7 Plurals

Some languages, such as English, have a general rule that plurals are formed from the singular with a suffix appended. This isn’t an absolute rule. There are plenty of exceptions (for example, geese, children, churches, elves, fairies, sheep). The glossaries package allows the plural key to be optional when defining entries. In some cases a plural may not make any sense (for example, the term is a symbol) and in some cases the plural may be identical to the singular.

To make life easier for languages where the majority of plurals can simply be formed by appending a suffix to the singular, the glossaries package lets the plural field default to the value of the text field with \glspluralsuffix appended. This command is defined to be just the letter “s”. This means that the majority of terms don’t need to have the plural supplied as well, and you only need to use it for the exceptions.

For languages that don’t have this general rule, the plural field will always need to be supplied, where needed.

There are other plural fields, such as firstplural, longplural and shortplural. Again, if you are using a language that doesn’t have a simple suffix rule, you’ll have to supply the plural forms if you need them (and if a plural makes sense in the context).

If these fields are omitted, the glossaries package follows these rules:

• If firstplural is missing, then \glspluralsuffix is appended to the first field, if that field has been supplied. If the first field hasn’t been supplied but the plural field has been supplied, then the firstplural field defaults to the plural field. If the plural field hasn’t been supplied, then both the plural and firstplural fields default to the text field (or name, if no text field) with \glspluralsuffix appended.
• If the longplural field is missing, then \glspluralsuffix is appended to the long field, if the long field has been supplied.
• If the shortplural field is missing then, with the base glossaries acronym mechanism, \acrpluralsuffix is appended to the short field.

This last case is changed with glossaries-extra. With this extension package, the shortplural field defaults to the short field with \abbrvpluralsuffix appended unless overridden by category attributes. This suffix command is set by the abbreviation styles. This means that every time an abbreviation style is implemented, \abbrvpluralsuffix is redefined. In most cases its redefined to use

---

---

which defaults to just \glspluralsuffix. Some of the abbreviation styles have their own command for the plural suffix, such as \glsxtrscsuffix, so if you want to completely strip all the plural suffixes used for abbreviations then you need to redefine \glsxtrabbrvpluralsuffix not \abbrvpluralsuffix, which changes with the style. Redefining \acrpluralsuffix will have no affect, since it’s not used by the new abbreviation mechanism.

If you require a mixture (for example, in a multilingual document), there are two attributes that affect the short plural suffix formation. The first is aposplural which uses the suffix

'\abbrvpluralsuffix

That is, an apostrophe followed by \abbrvpluralsuffix is appended. The second attribute is noshortplural which suppresses the suffix and simply sets shortplural to the same as short.

Top

2.8 Nested Links

Complications arise when you use \gls in the value of the name field (or text or first fields, if set). This tends to occur with abbreviations that extend other abbreviations. For example, SHTML is an abbreviation for SSI enabled HTML, where SSI is an abbreviation for Server Side Includes and HTML is an abbreviation for Hypertext Markup Language.

Things can go wrong if we try the following with the glossaries package:

\newacronym{ssi}{SSI}{Server Side Includes}
\newacronym{html}{HTML}{Hypertext Markup Language}
\newacronym{shtml}{S\gls{html}}{\gls{ssi} enabled \gls{html}}

The main problems are:

1. The first letter upper casing commands, such as \Gls, won’t work for the shtml entry on first use if the long form is displayed before the short form (which is the default abbreviation style). This will attempt to do

        \gls{\uppercase ssi} enabled \gls{html}
   

   which just doesn’t work. Grouping the \gls{ssi} doesn’t work either as this will effectively try to do

        \uppercase{\gls{ssi}} enabled \gls{html}
   

   This will upper case the label ssi so the entry won’t be recognised. This problem will also occur if you use the all capitals version, such as \GLS.

2. The long and abbreviated forms accessed through \glsentrylong and \glsentryshort are no longer expandable and so can’t be used be used in contexts that require this, such as PDF bookmarks.
3. The nested commands may end up in the sort key, which will confuse the indexing.
4. The shtml entry produces inconsistent results depending on whether the ssi or html entries have been used. Suppose both ssi and html are used before shtml. For example:

        This section discusses \gls{ssi}, \gls{html} and \gls{shtml}.
   

   This produces:

   This section discusses server side includes (SSI), hypertext markup language (HTML) and SSI enabled HTML (SHTML).

   So the first use of the shtml entry produces “SSI enabled HTML (SHTML)”.

   Now let’s suppose the html entry is used before the shtml but the ssi entry is used after the shtml entry, for example:

        The sample files are either \gls{html} or \gls{shtml}, but let's
        first discuss \gls{ssi}.
   

   This produces:

   The sample files are either hypertext markup language (HTML) or server side includes (SSI) enabled HTML (SHTML), but let’s first discuss SSI.

   So the first use of the shtml entry now produces “server side includes (SSI) enabled HTML (SHTML)”, which looks a bit strange.

   Now let’s suppose the shtml entry is used before (or without) the other two entries:

        This article is an introduction to \gls{shtml}.
   

   This produces:

   This article is an introduction to server side includes (SSI) enabled hypertext markup language (HTML) (SHTML).

   So the first use of the shtml entry now produces “server side includes (SSI) enabled hypertext markup language (HTML) (SHTML)”, which is even more strange.

   This is all aggravated by setting the style using the glossaries package’s \setacronymstyle. For example:

        \setacronymstyle{long-short}
   

   as this references the label through the use of \glslabel when displaying the long and short forms, but this value changes with each use of \gls, so instead of displaying “(SHTML)” at the end of the first use, it now displays “(HTML)”, since \glslabel has been changed to html by \gls{html}.

   Another oddity occurs if you reset the html entry between uses of the shtml entry. For example:

        \gls{shtml} ... \glsreset{html}\gls{shtml}
   

   The next use of shtml produces “Shypertext markup language (HTML)”, which is downright weird.

   Even without this, the short form has nested formatting commands, which amount to \acronymfont{S\acronymfont{HTML}}. This may not be a problem for some styles, but if you use one of the “sm” styles (that use \textsmaller), this will produce an odd result.

5. Each time the shtml entry is used, the html entry will also be indexed and marked as used, and on first use this will happen to both the ssi and html entries. This kind of duplication in the location list isn’t usually particularly helpful to the reader.
6. If hyperref is in use, you’ll get nested hyperlinks and there’s no consistent way of dealing with this across the available PDF viewers. If on the first use case, the user clicks on the “HTML” part of the “SSI enabled HTML (SHTML)” link, they may be directed to the HTML entry in the glossary or they may be directed to the SHTML entry in the glossary.

For these reasons it’s better to use the simple expandable commands like \glsentrytext or \glsentryshort in the definition of other entries (although that doesn’t fix the first problem).

Alternatively use something like:

\newacronym
 [description={\acrshort{ssi} enabled \acrshort{html}}]
 {shtml}{SHTML}{SSI enabled HTML}

with glossaries or:

\newabbreviation
 [description={\glsxtrshort{ssi} enabled \glsxtrshort{html}}]
 {shtml}{SHTML}{SSI enabled HTML}

with glossaries-extra. This fixes all the above listed problems (as long as you don’t use \glsdesc). Note that replacing \gls with \acrshort in the original example may fix the first use issue, but it doesn’t fix any of the other problems listed above.

If it’s simply that you want to use the abbreviation font, you can use \glsabbrvfont:

\setabbreviationstyle{long-short-sc}
\newabbreviation{ssi}{ssi}{server-side includes}
\newabbreviation{html}{html}{hypertext markup language}
\newabbreviation{shtml}{shtml}{\glsabbrvfont{ssi} enabled
\glsabbrvfont{html}}

This will pick up the font style setting of the outer entry (shtml, in the above case). This isn’t a problem in the above example as all the abbreviations use the same style.

If the term can simply be treated as a series of previously defined entries, then consider using multi-entries (or compound sets). See §6 Multi (or Compound) Entries for further details.

However, if you’re really determined to use \gls in a field that may be included within some link-text, glossaries-extra patches internals used by the linking commands so that if \gls (or plural or case changing variants) occurs in the link-text it will behave as though you used \glstext[hyper=false,noindex] instead. Grouping is also added so that, for example, when \gls{shtml} is used for the first time the long form

\gls{ssi} enabled \gls{html}

is treated as

{\glstext[hyper=false,noindex]{ssi}} enabled
{\glstext[hyper=false,noindex]{html}}

This overcomes problems 4, 5 and 6 listed above, but still doesn’t fix problems 1 and 2. Problem 3 usually won’t be an issue as most abbreviation styles set the sort key to the short form, so using these commands in the long form but not the short form will only affect entries with a style that sorts according to the long form (such as long-noshort-desc).

Additionally, any instance of the long form commands, such as \glsxtrlong or \acrlong will be temporarily redefined to just use

(or case-changing versions). Similarly the short form commands, such as \glsxtrshort or \acrshort will use \glsentryshort in the argument of either \glsabbrvfont (for \glsxtrshort) or \acronymfont (for \acrshort). So if the shtml entry had instead been defined as:

\newacronym{shtml}{SHTML}{\acrshort{ssi} enabled \acrshort{html}}

then (using the long-short style) the first use will be like

{\acronymfont{\glsentryshort{ssi}}} enabled
{\acronymfont{\glsentryshort{html}}} (SHTML)

whereas if the entry is defined as:

\newabbreviation{shtml}{SHTML}{\glsxtrshort{ssi} enabled
\glsxtrshort{html}}

then the first use will be like:

{\glsabbrvfont{\glsentryshort{ssi}}} enabled
{\glsabbrvfont{\glsentryshort{html}}} (SHTML)

Note that the first optional argument of \acrshort or \glsxtrshort is ignored in this context. (The final optional argument will be inserted, if present.) The abbreviation style that governs \glsabbrvfont will be set for \glsxtrshort. Note that \acrshort doesn’t set the abbreviation style.

Alternatively you can use:

---

---

where ⟨field⟩ is the field label and corresponds to a command in the form \gls⟨field⟩ (e.g. \glstext) or in the form \glsxtr⟨field⟩ (e.g. \glsxtrshort).

There’s a shortcut command for the most common fields:

---

---

which is equivalent to \glsxtrp{short}{⟨label⟩}, and

---

---

which is equivalent to \glsxtrp{text}{⟨label⟩}.

The \glsxtrp command behaves much like the \glsfmt⟨field⟩ commands described in §5 Entries in Sectioning Titles, Headers, Captions and Contents but the post-link hook is also suppressed and extra grouping is added. It automatically sets hyper to false and noindex to true. If you want to change this, you can use

---

---

For example:

\glsxtrsetpopts{hyper=false}

will just switch off the hyperlinks but not the indexing. Be careful using this command or you can end up back to the original problem of nested links.

The hyper link is re-enabled within glossaries. This is done through the command:

---

---

which by default just does

\glsxtrsetpopts{noindex}

You can redefine this if you want to adjust the setting when \glsxtrp is used in the glossary. For example:

\renewcommand{\glossxtrsetpopts}{\glsxtrsetpopts{noindex=false}}

For example,

\glsxtrp{short}{ssi}

is equivalent to

{\let\glspostlinkhook\relax
 \glsxtrshort[hyper=false,noindex]{ssi}[]%
}

in the main body of the document or

{\let\glspostlinkhook\relax
 \glsxtrshort[noindex]{ssi}[]%
}

inside the glossary. (Note the post-link hook is locally disabled.)

If \glsxtrp{short}{ssi} occurs in a sectioning mark, it’s equivalent to

{\glsxtrheadshort{ssi}}

(which recognises the headuc attribute.)

If hyperref has been loaded, then the bookmark will use \glsentry⟨field⟩ (\glsentryshort{ssi} in the above example).

There are similar commands

---

---

for first letter upper case and

---

---

for all upper case.

You can, with care, protect against issue 1 by inserting an empty group at the start if the long form starts with a command that breaks the first letter uppercasing commands like \Gls, but you still won’t be able to use the all caps commands, such as \GLS.

If you really need nested commands, the safest method is

\newabbreviation{shtml}{shtml}{{}\glsxtrp{short}{ssi} enabled
\glsxtrp{short}{html}}

but be aware that it may have some unexpected results occasionally.

Example document:

\documentclass{report}
\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{slantsc}
\usepackage[colorlinks]{hyperref}
\usepackage[nopostdot=false]{glossaries-extra}
\makeglossaries
\setabbreviationstyle{long-short-sc}
\newabbreviation{ssi}{ssi}{server-side includes}
\newabbreviation{html}{html}{hypertext markup language}
\newabbreviation{shtml}{shtml}{{}\glsps{ssi} enabled {}\glsps{html}}
\pagestyle{headings}
\glssetcategoryattribute{abbreviation}{headuc}{true}
\glssetcategoryattribute{abbreviation}{glossdesc}{title}
\begin{document}
\tableofcontents
\chapter{\glsfmtfull{shtml}}
First use: \gls{shtml}, \gls{ssi} and \gls{html}.
Next use: \gls{shtml}, \gls{ssi} and \gls{html}.
\newpage
Next page.
\printglossaries
\end{document}

Top

2.9 Acronym Style Modifications

The glossaries-extra package provides a new way of dealing with abbreviations and redefines \newacronym to use \newabbreviation (see §4 Abbreviations). The simplest way to update a document that uses \newacronym from glossaries to glossaries-extra is do just add

\setabbreviationstyle[acronym]{long-short}

before you define any entries. If you have used commands like \acrshort, \acrlong and \acrfull, you need to substitute them with \glsxtrshort, \glsxtrlong and \glsxtrfull. (This includes plural and case-changing forms.) Your text editor’s search and replace function should help with this. If you have used the shortcuts package option then use shortcuts=ac in order to continue using the shortcut commands like \ac. With this setting, \acs will use \glsxtrshort instead of \acrshort etc.

For example, the following document using just glossaries

\documentclass{article}
\usepackage[acronym,nopostdot,toc]{glossaries}
\makeglossaries
\setacronymstyle{long-short}
\newacronym{html}{HTML}{hypertext markup language}
\begin{document}
\gls{html}
\printglossaries
\end{document}

can be easily adapted to use glossaries-extra:

\documentclass{article}
\usepackage[acronym]{glossaries-extra}
\makeglossaries
\setabbreviationstyle[acronym]{long-short}
\newacronym{html}{HTML}{hypertext markup language}
\begin{document}
\gls{html}
\printglossaries
\end{document}

Table 2.1 lists the nearest equivalent glossaries-extra abbreviation styles for the predefined acronym styles provided by glossaries, but note that the new styles use different formatting commands. See §4.4 Predefined Abbreviation Styles for further details.

The reason for introducing the new style of abbreviation commands provided by glossaries-extra is because the original acronym commands provided by glossaries are too restrictive to work with the internal modifications made by glossaries-extra. However, if you really want to restore the generic acronym function provided by glossaries you can use

---

---

(before any use of \newacronym).

\RestoreAcronyms should not be used in combination with the newer glossaries-extra abbreviations. Don’t combine old and new style entries with the same type. The original glossaries acronym mechanism doesn’t work well with the newer glossaries-extra commands.

In general, there’s rarely any need for \RestoreAcronyms. If you have a document that uses \newacronymstyle, then it’s best to either stick with just glossaries for that document or define an equivalent abbreviation style with \newabbreviationstyle. (See §4.5 Defining New Abbreviation Styles for further details.)

---

---

The space command \glsacspace used by the long-sp-short acronym style provided by glossaries is modified so that it uses

---

---

instead of the hard-coded 3em. This is a command not a length and so can be changed using \renewcommand.

Any of the new abbreviation styles that use \glsxtrfullsep (such as long-short) can easily be changed to use \glsacspace with

\renewcommand*{\glsxtrfullsep}[1]{\glsacspace{#1}}

The first use acronym font command

---

---

is redefined to use the first use abbreviation font command \glsfirstabbrvfont. This will be reset if you use \RestoreAcronyms.

The subsequent use acronym font command

---

---

is redefined to use the subsequent use abbreviation font command \glsabbrvfont. This will be reset if you use \RestoreAcronyms.

Top

2.10 Glossaries

The \newignoredglossary{⟨type⟩} command now (as from v1.11) has a starred version that doesn’t automatically switch off the hyperlinks. This starred version may be used with the targeturl attribute to create a link to an external URL. (See §7 Categories for further details.) As from v1.12 both the starred and unstarred version check that the glossary doesn’t already exist. (The glossaries package omits this check.)

You can now provide an ignored glossary with:

---

---

which will only define the glossary if it doesn’t already exist. This also has a starred version that doesn’t automatically switch off hyperlinks.

The individual glossary displaying commands \printglossary, \printnoidxglossary and \printunsrtglossary have extra keys:

• target. This is a boolean key which can be used to switch off the automatic hypertarget for each entry. Unlike \glsdisablehyper this doesn’t switch off hyperlinks, so any cross-references within the glossary won’t be affected. This is a way of avoiding duplicate target warnings if a glossary needs to be displayed multiple times.
• targetnameprefix={⟨prefix⟩}. Another way of avoiding duplicate target names is to set a prefix used for the names. Unlike changing \glolinkprefix this doesn’t affect any hyperlinks (such as those created with \gls).
  This doesn’t include the navigation hyper group targets in styles such as listhypergroup. The link name for these is given by \glsnavhyperlinkname.

• prefix={⟨prefix⟩}. If you do actually want to locally change \glolinkprefix, you can use the prefix key instead. You need to use the matching prefix key in commands like \gls.
• label={⟨label⟩}. Labels the glossary with \label{⟨label⟩} and also defines \@currentlabelname to the glossary’s TOC title (to provide text for \nameref). You can also set the label before the glossary using:

  ---

  \glsxtrsetglossarylabel{⟨label⟩}

  ---

  but take care of duplicate labels if it’s not scoped.

• groups. This is a boolean key (new to v1.44) designed for use with \printunsrtglossary (and other \printunsrt… variations). It has no effect with \printglossary and \printnoidxglossary.

  If true (default), group formation (group header and group skip) will be attempted if the group key has been defined and set. Note that some styles ignore the header and group skip commands so there may not be a noticeable difference in those cases. If this key is set to false, no group formation will occur so there won’t be any group markup to separate letter groups so there will be no visual separation regardless of style or the group skip setting.

• leveloffset. This key (new to v1.44) is designed for use with \printunsrtglossary (and other \printunsrt… variations). It has no effect with \printglossary and \printnoidxglossary.

  The value may either be a simple integer ⟨n⟩ to indicate assignment or in the form ++⟨n⟩ to indicate an increment. This will locally assign or increment the level offset. The default value is 0. This will cause entries to be displayed as though their hierarchical level is N more than it would normally be (where N is the level offset). For example, a top-level entry (that is, an entry without a parent) would normally have a hierarchical level of 0. With a level offset of 1, it would be treated by the glossary style as though it was actually a level 1 entry. (Remember that some styles don’t support hierarchical glossaries.)

  Note that the group formation (if supported) will still occur between entries that don’t have a parent, regardless of the level offset. This can cause odd results but can be switched off with groups. See Gallery: Inner or Nested Glossaries for an example.

Top

2.11 Glossary Style Modifications

The glossaries-extra-stylemods package (more conveniently loaded through the glossaries-extra stylemods option) modifies some of the predefined styles that are provided with the glossaries package. These modifications are described in more detail in §2.11.3 The glossaries-extra-stylemods Package.

The glossaries package tries to determine the group title from its label by first checking if \⟨label⟩groupname exists. If it doesn’t exist, then the title is assumed to be the same as the label. For example, when typesetting the “A” letter group, glossaries first checks if \Agroupname exists. This could potentially cause conflict with another package that may have some other meaning for \Agroupname, so glossaries-extra first checks for the existence of the internal command \glsxtr@grouptitle@⟨label⟩ which shouldn’t clash with another package. You can set the group title using

---

---

For example:

\glsxtrsetgrouptitle{A}{A (a)}

This uses a global assignment. If you need to scope the change you can use

---

---

Top

2.11.1 Style Hooks

The commands \glossentryname and \glossentrydesc are modified to take into account the glossname, glossnamefont, glossdesc and glossdescfont attributes (see §7 Categories). This means you can make simple font or case-changing modifications to the name and description without defining a new glossary style.

(New to version 1.42.) The command \glossentrysymbol is modified to take into account the glosssymbolfont attribute. Note that, unlike the above, there’s no corresponding attribute to change the case as it’s usually not appropriate to change the case of a symbol (and for some symbols, such as pictographs, there’s no concept of case). If \texorpdfstring has been defined \glossentrysymbol will be defined to do:

The ⟨TeX code⟩ part is robust and deals with the actual typesetting of the symbol. The ⟨PDF⟩ part is simply:

---

---

which is defined to just do \glsentrysymbol{⟨label⟩}. The chances are that the code in the symbol key won’t be valid in the PDF bookmarks, so you can redefine \glsentrypdfsymbol to use a more appropriate field. (If you do redefine this command, remember that it needs to fully expand.)

For example, if you are using glossaries-accsupp, you could use the symbolaccess field:

\renewcommand{\glsentrypdfsymbol}[1]{\glsentrysymbolaccess{#1}}

Alternatively, if you are using bib2gls you can use the TeX parser library to interpret a copy of the symbol field and use that. For example, with the resource options:

  replicate-fields={symbol=user1},
  interpret-fields={user1}

This stores the interpreted value of the symbol in the user1 field, so you can then do:

\renewcommand{\glsentrypdfsymbol}[1]{\glsentryuseri{#1}}

(You may need XƎLATEX or LuaLATEX with this method.) This allows \glossentrysymbol to be used in a section heading with standalone definitions.

If you want to adapt a style to use another field instead of name, you can use

---

---

This behaves just like \glossentryname (that is, it obeys the glossname attribute, uses either the glossnamefont attribute or \glsnamefont to format the text, and uses the post-name hook) but it uses the given ⟨field⟩ instead of name. The ⟨field⟩ argument must be the internal field name (for example desc rather than description). See the key to field mappings table in the glossaries user manual.

There is a hook after \glossentryname and \Glossentryname:

---

---

By default this checks the indexname attribute. If the attribute exists for the category to which the label belongs, then the name is automatically indexed using

---

---

See §9 Auto-Indexing for further details.

As from version 1.04, the post-name hook \glsxtrpostnamehook will also use \glsxtrpostname⟨category⟩ if it exists. You can use \glscurrententrylabel to obtain the entry label with the definition of this command. For example, suppose you are using a glossary style the doesn’t display the symbol, you can insert the symbol after the name for a particular category, say, the “symbol” category:

\newcommand*{\glsxtrpostnamesymbol}{\space
 (\glsentrysymbol{\glscurrententrylabel})}

For convenience, as from v1.31, you can use

---

---

This is simply a shortcut for:

Note that it doesn’t check if the command has already been defined.

As from version 1.25, the post-name hook also does

---

---

(before \glsxtrpostname⟨category⟩) to allow for additional non-category related code. This does nothing by default.

The post-description code used within the glossary is modified so that it also does

---

---

This occurs before the original \glspostdescription, so if the nopostdot=false option is used, it will be inserted before the terminating full stop.

This new command will do \glsxtrpostdesc⟨category⟩ if it exists, where ⟨category⟩ is the category label associated with the current entry. For example \glsxtrpostdescgeneral for entries with the category set to general or \glsxtrpostdescacronym for entries with the category set to acronym. For convenience, as from v1.31, you can use

---

---

This is simply a shortcut for:

Note that it doesn’t check if the command has already been defined.

Since both \glossentry and \subglossentry set

---

---

to the label for the current entry, you can use this within the definition of these post-description hooks if you need to reference the label.

For example, suppose you want to insert the plural form in brackets after the description in the glossary, but only for entries in the general category, then you could do:

\renewcommand{\glsxtrpostdescgeneral}{\space
 (plural: \glsentryplural{\glscurrententrylabel})}

This means you don’t have to define a custom glossary style, which you may find more complicated. (It also allows more flexibility if you decide to change the underlying glossary style.)

Top

2.11.2 Number List

The number list is now placed inside the argument of

---

---

This is internally used by \glossaryentrynumbers. The nonumberlist option redefines \glossaryentrynumbers so that it doesn’t display the number list, but it still saves the number list in case it’s required.

For example, to change the font for the entire number list redefine \GlsXtrFormatLocationList as appropriate. Don’t modify \glossaryentrynumbers.

Sometimes users like to insert “page” or “pages” in front of the number list. This is quite fiddly to do with the base glossaries package, but glossaries-extra provides a way of doing this. First you need to enable this option and specify the text to display using:

---

---

where ⟨page⟩ is the text to display if the number list only contains a single location and ⟨pages⟩ is the text to display otherwise. For example:

\GlsXtrEnablePreLocationTag{Page: }{Pages: }

An extra run is required when using this command.

See the accompanying sample file sample-pages.tex.

Note that bib2gls can be instructed to insert a prefix at the start of non-empty location lists, which can be used as an alternative to \GlsXtrEnablePreLocationTag.

Location lists displayed with \printnoidxglossary internally use

---

---

This command is provided by glossaries, but is modified by glossaries-extra to check for the start and end range formation identifiers ( and ) which are discarded to obtain the actual control sequence name that forms the location formatting command.

If the range identifiers aren’t present, this just uses

---

---

otherwise it uses

---

---

for the start of a range (where the identifier has been stripped from ⟨format⟩) or

---

---

for the end of a range (where the identifier has been stripped from ⟨format⟩).

By default the start range command saves the format in

---

---

and does
\glsxtrdisplaysingleloc{⟨format⟩}{⟨location⟩}
(If the format is empty, it will be replaced with glsnumberformat.)

The end command checks that the format matches the start of the range, does

---

---

(which does nothing by default), followed by
\glsxtrdisplaysingleloc{⟨format⟩}{⟨location⟩}
and then sets \glsxtrlocrangefmt to empty.

This means that the list

\glsnoidxdisplayloc{}{page}{(textbf}{1},
\glsnoidxdisplayloc{}{page}{textbf}{1},
\glsnoidxdisplayloc{}{page}{)textbf}{1}.

doesn’t display any differently from

\glsnoidxdisplayloc{}{page}{textbf}{1},
\glsnoidxdisplayloc{}{page}{textbf}{1},
\glsnoidxdisplayloc{}{page}{textbf}{1}.

but it does make it easier to define your own custom list handler that can accommodate the ranges.

Top

2.11.3 The glossaries-extra-stylemods Package

As from v1.02, glossaries-extra now includes the package glossaries-extra-stylemods that will redefine the predefined styles to include the post-description hook (for those that are missing it). You will need to make sure the styles have already been defined before loading glossaries-extra. For example:

\usepackage{glossaries-extra}
\usepackage{glossary-longragged}
\usepackage{glossaries-extra-stylemods}

Alternatively you can load glossary-⟨name⟩.sty at the same time by passing ⟨name⟩ as a package option to glossaries-extra-stylemods. For example:

\usepackage{glossaries-extra}
\usepackage[longragged]{glossaries-extra-stylemods}

Another option is to use the stylemods key when you load glossaries-extra. You can omit a value if you only want to use the predefined styles that are automatically loaded by glossaries (for example, the long3col style):

\usepackage[style=long3col,stylemods]{glossaries-extra}

Or the value of stylemods may be a comma-separated list of the style package identifiers. For example:

\usepackage[style=mcoltree,stylemods=mcols]{glossaries-extra}

Remember to group the value if it contains any commas:

\usepackage[stylemods={mcols,longbooktabs}]{glossaries-extra}

Inline Style Note that the inline style is dealt with slightly differently. The original definition provided by the glossary-inline package uses \glspostdescription at the end of the glossary (not after each entry description) within the definition of \glspostinline. The style modification changes this so that \glspostinline just does a full stop followed by space factor adjustment, and the description \glsinlinedescformat and sub-entry description formats \glsinlinesubdescformat are redefined to include \glsxtrpostdescription (not \glspostdescription). This means that the modified inline style isn’t affected by the nopostdot option, but the post-description category hook can still be used.

Tabular Styles The tabular-like styles, such as long are adjusted so that the \ifglsnogroupskip conditional (set with nogroupskip) is moved outside of the definition of \glsgroupskip to avoid problems that cause an “Incomplete \iftrue” error with \printunsrtglossary and \printnoidxglossary. This means that if you want to change this conditional using \setupglossaries or using the nogroupskip option in \printglossary, \printnoidxglossary or \printunsrtglossary, you must also reset the glossary style.

As from version 1.21, the hard-coded \space before the number list in many of the predefined styles is replaced with

---

---

This just defaults to \space but may be redefined as required. For example:

\renewcommand{\glsxtrprelocation}{\dotfill}

List Styles The list styles use

---

---

(which defaults to \glsxtrprelocation) for top-level items and

---

---

(which defaults to \glslistprelocation) for child items.

As from v1.31, the description (including the post-description hook) is governed by:

---

---

for the list and altlist styles (but not the listdotted variations).

As from v1.47, the item is governed by:

---

---

The altlist styles use:

---

---

which internally uses \glslistitem. The header item (for the list styles that should the group title) is governed by:

---

---

This ignores the ⟨group-label⟩ by default and simply places the second argument in the optional argument of \item. The ⟨header code⟩ is the formatted group title, possibly including a hypertarget. The spacing after the group item is given by:

---

---

For just the list style and its letter group variations (not the altlist or listdotted variations) the number list for child entries is followed by

---

---

which defaults to a full stop.

The default value of \glslistdottedwidth is changed so that it’s set at the start of the document (if it hasn’t been changed in the preamble). This should take into account situations where \hsize isn’t set until the start of the document.

The separator between groups (if not nogroupskip) is now given by:

---

---

This defaults to \indexspace with penalties to deter page breaks. This command isn’t used if nogroupskip is set.

The glossary-tree package introduced new commands in v4.22, \glstreegroupheaderfmt and \glstreenavigationfmt, which are used to format the letter group headings and the navigation elements for the appropriate styles. These two new commands are defined in terms of \glstreenamefmt since that was the command originally used for the group headings and navigation. This now allows these different elements to be defined independently, but the most common redefinition is for \glstreenamefmt to remove the bold in the name. If the bold is still required for the group heading and navigation elements, then both other commands also need redefining. To simplify matters, if \glstreenamefmt has been defined, as from v1.31 glossaries-extra-stylemods defines:

---

---

which simply does \textbf{⟨text⟩} and redefines \glstreenamefmt, \glstreegroupheaderfmt and \glstreenavigationfmt all in terms of \glstreedefaultnamefmt.

This means that if you want to change all three to use a particular style you only need to redefine \glstreedefaultnamefmt, but if you only want to redefine \glstreenamefmt without affecting the other two commands, then you now can.

The separator between groups without headers is given by:

---

---

This defaults to just \indexspace without penalties. This command isn’t used if nogroupskip is set. (The penalties introduced in v1.41 were moved to \glstreeheadergroupskip in v1.42 as they are inappropriate when there’s no header.)

The separator between groups with headers is now given by (as from v1.42):

---

---

This defaults to \glstreegroupskip with penalties to deter page breaks after the group heading.

The styles that display the group titles now use:

---

---

where ⟨label⟩ is the group label and ⟨title⟩ is the group title. This does nothing by default and is inserted before the group title. You can redefine it to add the group title to the PDF bookmarks. For example, if the glossary title uses \chapter then:

\renewcommand{\glstreePreHeader}[2]{%
 \pdfbookmark[1]{#2}{\currentglossary.#1}%
}

will insert section-level bookmarks. The use of \currentglossary helps to provide unique bookmark labels in the event of multiple glossaries.

The glossary-tree package provides the commands

---

---

and

---

---

(which both default to a space) and uses them in the tree-like styles, but not for the alttree style. The glossaries-extra-stylemods package modifies the alttree style so that it has equivalent hooks:

---

---

and

---

---

These do nothing by default.

The index-like and tree-like styles insert the pre-number list space with

---

---

(which defaults to \glsxtrprelocation) for top-level items and

---

---

(which defaults to \glstreeprelocation) for child items.

As from version 1.31, the glossaries-extra-stylemods package also provides:

---

---

which is used by the treenoname styles to display the pre-description separator, the description and the post-description hook. Similarly for the symbol:

---

---

The above are just used for top-level entries. Child entries don’t have the name or symbol displayed for the treenoname styles, so there’s only a command for the child description:

---

---

For the tree styles (but not the treenoname or alttree styles), the description is displayed using:

---

---

and the symbol with:

---

---

Again the above two commands are just for top-level entries. The child entries use:

---

---

for the description and

---

---

for the symbol. As from version 1.41, there are now wrapper commands for \glstreedesc and \glstreechilddesc that check for the description and symbol to determine what separator to use before the page list. These are:

---

---

for top-level entries and

---

---

for sub-entries.

If either the symbol or description is present these will use \glstreeprelocation or \glstreechildprelocation, respectively. Otherwise, both will use (from v1.42):

---

---

The default is a space. This means that you could have, say, a comma followed by a space for terms that are simply an alias, but just have a space for terms that have a description that ends with a full stop (or that just have a symbol without a description) where the comma would be inappropriate.

Note that version 1.42 has corrected an error that was introduced to v1.41 that caused the name to run into the location list if there was no symbol and no description.

As from version 1.05, the glossaries-extra-stylemods package provides some additional commands for use with the alttree style to make it easier to modify. These commands are only defined if the glossary-tree package has already been loaded, which is typically the case unless the notree option has been used when loading glossaries.

---

---

(New to version 1.21.) This is like \glssetwidest (provided by glossary-tree) but performs a global assignment.

---

---

This is like \glssetwidest but performs a protected expansion on ⟨name⟩. This has a localised effect. For a global setting, use

---

---

The following only set the value if ⟨name⟩ is wider than the current value (new to version 1.23). Local update:

---

---

Global update:

---

---

Locale update (expands ⟨name⟩):

---

---

Global update (expands ⟨name⟩):

---

---

The widest entry value can later be retrieved using

---

---

for the top-level entries and

---

---

for sub-entries, where ⟨level⟩ is the level number.

Note that if you are using bib2gls, you can use the resource option set-widest which will try to determine the widest name of all the selected entries. This isn’t guaranteed to work as it may depend on fonts or commands that bib2gls can’t replicate, but it should be suitable for names that just consist of text, and can be more efficient than iterating over all the defined entries using TeX.

The command \glsfindwidesttoplevelname provided by glossary-tree has a CamelCase synonym:

---

---

Similar commands are also provided:

---

---

This has an additional check that the entry has been used. Naturally this is only useful if the glossaries that use the alttree style occur at the end of the document. This command should be placed just before the start of the glossary. (Alternatively, place it at the end of the document and save the value in the auxiliary file for the next run.)

---

---

This is like the previous command but if doesn’t check the parent key. This is useful if all levels should have the same width for the name.

---

---

This is like the previous command but doesn’t check if the entry has been used.

---

---

This is like \glsFindWidestUsedTopLevelName but also sets the first two sub-levels as well. Any entry that has a great-grandparent is ignored.

---

---

This is like the previous command but doesn’t check if the entry has been used.

---

---

This is like \glsFindWidestUsedAnyName but also measures the symbol. The length of the widest symbol is stored in ⟨register⟩.

---

---

This is like the previous command but it doesn’t check if the entry has been used.

---

---

This is like \glsFindWidestUsedAnyNameSymbol but also measures the number list. This requires \glsentrynumberlist (see the glossaries user manual). The length of the widest symbol is stored in ⟨symbol register⟩ and the length of the widest number list is stored in ⟨location register⟩.

---

---

This is like the previous command but it doesn’t check if the entry has been used.

---

---

This is like \glsFindWidestUsedAnyNameSymbolLocation but doesn’t measure the symbol. The length of the widest number list is stored in ⟨register⟩.

---

---

This is like the previous command but doesn’t check if the entry has been used.

The layout of the symbol, description and number list is governed by

---

---

for top-level entries and

---

---

for sub-entries.

There is now a user level command that performs the initialisation for the alttree style:

---

---

The paragraph indent for subsequent paragraphs in multi-paragraph descriptions is provided by the length

---

---

For additional commands that are available with the alttree style, see the documented code (glossaries-extra-code.pdf). See also the accompanying sample files sample-alttree.tex, sample-alttree-sym.tex and sample-alttree-marginpar.tex.

3. New Glossary Styles

The glossaries-extra package comes with some new styles. The associated style package needs to be loaded. This can be done with \usepackage but it’s simpler to use the stylemods option.

Top

3.1 glossary-bookindex package

As from v1.21, glossaries-extra has a new supplementary package glossary-bookindex which provides the glossary style bookindex. This is very similar to the mcolindexgroup style but is designed for indexes, so by default only the name and location list are displayed. You can either load this package explicitly and then set the style:

\usepackage{glossaries-extra}
\usepackage{glossary-bookindex}
\setglossarystyle{bookindex}

or use both the stylemods and style options:

\usepackage[stylemods=bookindex,style=bookindex]{glossaries-extra}

The bookindex style only supports a maximum hierarchical level of 2 (top-level, level 1 and level 2). It’s primarily designed for use with bib2gls. It may be used with other indexing options, but some features may not be present and UTF-8 characters may cause a problem with non-Unicode engines in letter group headings or PDF bookmarks. (bib2gls uses numeric identifies by default to avoid these problems. If numbers show up in the group headings instead of the appropriate characters, check that you have used the record package option.)

The number of columns is given by

---

---

which defaults to 2.

This style uses the multicols environment. If the command

---

---

isn’t empty then it’s supplied as the optional argument following \begin{multicols} {⟨n⟩}. You can switch from multicols to multicols* by redefining

---

---

For example

\renewcommand{\glsxtrbookindexmulticolsenv}{multicols*}

Each top-level entry is displayed using

---

---

where the entry is identified by ⟨label⟩. This just does \glossentryname{⟨label⟩} by default. For example, if you want the symbol to be included:

\renewcommand*{\glsxtrbookindexname}[1]{%
 \glossentryname{#1}%
 \ifglshassymbol{#1}{\space (\glossentrysymbol{#1})}{}%
}

or if you want the description (if set):

\renewcommand*{\glsxtrbookindexname}[1]{%
 \glossentryname{#1}%
 \ifglshasdesc{#1}{\space \glossentrydesc{#1}\glspostdescription}{}%
}

(which picks up the post-description hook).

Alternatively you can use the \glsxtrpostname⟨category⟩ hook to append information after the name according to the entry’s category.

Sub-entries are displayed using

---

---

which just defaults to \glsxtrbookindexname{⟨label⟩}.

The separator used before the location list for top-level entries is given by

---

---

where ⟨label⟩ is the entry’s label. This checks if the location field has been set. If it has, it does

,\glsxtrprelocation

otherwise it just does \glsxtrprelocation (which defaults to \space). If you’re not using bib2gls, the location field won’t be set.

The separator used before the location list for sub-entries is given by

---

---

which defaults to \glsxtrbookindexprelocation{⟨label⟩}.

The actual location list is encapsulated with:

---

---

for top-level entries and

---

---

for sub-entries. These both just do ⟨location list⟩ by default.

The separator used between a top-level parent and child entry is given by

---

---

This defaults to \nopagebreak.

The separator used between a sub-level parent and child entry is given by

---

---

This defaults to \glsxtrbookindexparentchildsep.

The separator between top-level entries is given by

---

---

This comes after the entry given by ⟨label1⟩, if the entry has no children, or after the last descendent otherwise, so it always comes immediately before the entry given by ⟨label2⟩ unless the entry occurs at the start of a group. This does nothing by default.

The separator between two level 1 entries is given by

---

---

The separator between two level 2 entries is given by

---

---

At the end of each letter group, the following hooks are done in order:

---

---

---

---

---

---

where ⟨sub-sub-label⟩ is the label of the last level 2 entry, ⟨sub-label⟩ is the label of the last level 1 entry and ⟨label⟩ is the label of the last level 0 entry.

For example, the resource option seealso=omit instructs bib2gls to omit the seealso cross-reference from the location list. (The see cross-reference will still be added unless you also have see=omit.) The seealso cross-reference can instead be appended after the child entries using:

\renewcommand{\glsxtrbookindexatendgroup}[1]{%
  \glsxtrifhasfield{seealso}{#1}{\glstreesubitem\glsxtruseseealso{#1}}{}%
}
\renewcommand{\glsxtrbookindexbetween}[2]{%
  \glsxtrbookindexatendgroup{#1}%
}%
\renewcommand{\glsxtrbookindexsubatendgroup}[1]{%
  \glsxtrifhasfield{seealso}{#1}{\glstreesubsubitem\glsxtruseseealso{#1}}{}%
}
\renewcommand{\glsxtrbookindexsubbetween}[2]{%
  \glsxtrbookindexsubatendgroup{#1}%
}
\renewcommand{\glsxtrbookindexsubsubatendgroup}[1]{%
  \glsxtrifhasfield{seealso}{#1}%
  {\glstreeitem\hspace*{40pt}\glsxtruseseealso{#1}}{}%
}
\renewcommand{\glsxtrbookindexsubsubbetween}[2]{%
  \glsxtrbookindexsubsubatendgroup{#1}%
}

This uses \glstreesubitem and \glstreesubsubitem to indent the cross-reference according to the next level down, so the cross-reference for a top-level entry is aligned with the sub-entries, and a level 1 entry has its cross-reference aligned with sub-sub-entries. In the event that a level 2 entry has a cross-reference, this is indented a bit further (but it won’t be aligned with any deeper level as the bookindex style only supports a maximum of two sub-levels).

The bookindex style uses group headings. (If you use bib2gls remember to invoke it with the --group or -g switch.) The heading will use

---

---

If \pdfbookmark has been defined, this will use that command to bookmark the group title. If section=chapter is set (default if chapters are defined) then this uses level 1 otherwise it uses level 2. You can redefine this command if this isn’t appropriate. If \pdfbookmark hasn’t been defined, this command does nothin.

The group heading is formatted according to

---

---

which is defined as

\newcommand*{\glsxtrbookindexformatheader}[1]{%
  \par{\centering\glstreegroupheaderfmt{#1}\par}%
}

where \glstreegroupheaderfmt is provided by the glossary-tree package, which is automatically loaded. Note that the entry names aren’t encapsulated with \glstreenamefmt.

The glossary-bookindex package provides some supplementary commands that aren’t used by default, but may be used when adjusting the style. These commands should only be used within one of the \print…glossary commands. (That is, they should only be used in glossary styles.)

---

---

This writes information to the .aux file that can be read on the next run to obtain the first and last entry on each page of the glossary.

You can display the first entry associated with the current page using:

---

---

and the last entry associated with the current page using:

---

---

These do nothing if there are no entries marked on the current page (or if the document build isn’t up to date).

The entry is formatted using:

---

---

for the first instance and

---

---

for the last.

These commands are designed for use in page headers or footers where the page number is stable. For example, \glsxtrbookindexname can be redefined to mark the current entry:

\renewcommand{\glsxtrbookindexname}[1]{%
  \glsxtrbookindexmarkentry{#1}%
  \glossentryname{#1}%
}

If you only want to mark the top-level entries, remember to redefine \glsxtrbookindexsubname as it defaults to \glsxtrbookindexname:

\renewcommand{\glsxtrbookindexsubname}[1]{%
  \glossentryname{#1}%
}

Then if you’re using fancyhdr you can set the page style to show the first and last entry for the current page with:

  \pagestyle{fancy}%
  \lhead{\thepage}%
  \lfoot{\glsxtrbookindexfirstmark}%
  \cfoot{}%
  \rfoot{\glsxtrbookindexlastmark}%

Top

3.2 glossary-longextra package

As from version 1.37, the glossaries-extra package comes with the supplementary package glossary-longextra that provides additional styles, listed below, that use the longtable environment. If you know that your glossary won’t span more than a page and you need to use it in a context that’s incompatible with longtable, you can instead setup these styles to use tabular instead. In order to do this you must use

---

---

before the style is set. For example:

\GlsLongExtraUseTabulartrue
\setglossarystyle{long-name-desc}

or

\GlsLongExtraUseTabulartrue
\printunsrtglossary[style={long-name-desc}]

If you use this setting, you can change the default vertical alignment with:

---

---

The default definition is c.

The column titles are formatted according to:

---

---

which simply does \textbf{⟨text⟩} by default.

The name column has the title given by \entryname and the column alignment is given by:

---

---

which expands to l (left) by default.

The symbol column (where applicable) has the title given by \symbolname and the column alignment is given by:

---

---

which expands to c (centred) by default.

The location list column (where applicable) has the title given by \pagelistname and the column alignment is given by:

---

---

which expands to

>{\raggedright}p{\glspagelistwidth}

by default. (Ragged-right paragraph, \glspagelistwidth is defined in glossary-long, which is automatically loaded.)

The description column has the title given by \descriptionname and the column alignment is given by:

---

---

which expands to

>{\raggedright}p{\glsdescwidth}

by default. (Ragged-right paragraph, \glsdescwidth is defined in glossary-long, which is automatically loaded.)

Unlike the long styles provided by the base glossaries package, these new styles try to determine the value of \glsdescwidth at the start of the glossary according to the number of columns provided by the style. The calculations are performed by the following commands:

---

---

This is used by the styles that have a name and description column. The value of \glsdescwidth is computed as:

where W is a guess at the width of the name column. This is first set to the width of the name column header:

If any names in that column are larger than this, then you need to specify the widest name using:

---

---

or

---

---

These work like the analogous commands \glssetwidest and \glsupdatewidest provided with the alttree style, but in this case there’s no hierarchy. The default widest name is obtained from the alttree top-level name if set, otherwise it’s empty, so you can use bib2gls’s set-widest option. If you have the entry counter enabled, you will need to include this with the name for the extra material to be taken into account.

The name isn’t shown for child entries by default, but if you change this and you want to use bib2gls’s set-widest option (for bib2gls v1.8+) then you need to redefine:

---

---

This does nothing by default, but if you are including the child names then you need to redefine this command:

\renewcommand{\glslongextraUpdateWidestChild}[2]{%
 \glslongextraUpdateWidest{#2}%
}

If you prefer to set an explicit width for the description column then you need to redefine \glslongextraSetDescWidth. For example:

\renewcommand{\glslongextraSetDescWidth}{%
  \setlength{\glsdescwidth}{0.6\linewidth}%
}

The styles that have a name, symbol and description, \glsdescwidth is set with:

---

---

This first uses \glslongextraSetDescWidth and then subtracts 2\tabcolsep and the width of the symbol column header from \glsdescwidth. This assumes that the symbol column header is larger than any of the symbols. If this isn’t appropriate then you can redefine this command. For example:

\renewcommand{\glslongextraSymSetDescWidth}{%
  \glslongextraSetDescWidth
  \addtolength{\glsdescwidth}{-3cm}%
}

or

\renewcommand{\glslongextraSymSetDescWidth}{%
  \setlength{\glsdescwidth}{.5\linewidth}%
}

For the styles that have a name, description and location column, \glsdescwidth is set using:

---

---

This uses \glslongextraSetDescWidth and then subtracts 2\tabcolsep and \glspagelistwidth from \glsdescwidth. You can redefine this command to set both \glsdescwidth and \glspagelistwidth if appropriate.

For the styles that have a name, description, symbol and location column, \glsdescwidth is set using:

---

---

This uses \glslongextraSymSetDescWidth and then subtracts 2\tabcolsep and \glspagelistwidth from \glsdescwidth. Again, you can redefine this command to explicitly set both lengths.

In all cases, the top-level name is formatted according to:

---

---

This does

which includes the entry counter (if enabled), the target and the post-name link.

The top-level description is formatted according to:

---

---

This does \glossentrydesc{⟨label⟩} followed by the post-description hook.

The styles that have a symbol column format the symbol using:

---

---

This just does \glossentrysymbol{⟨label⟩}.

The styles that have a location list column format the list using:

---

---

This just does ⟨locations⟩ and ignores the label.

The child entries have their name formatted according to:

---

---

where ⟨level⟩ is the hierarchical level. This defaults to:

This includes the sub-entry counter (if enabled) and the target but doesn’t show the name. The child description is formatted according to:

---

---

which defaults to just \glslongextraDescFmt{⟨label⟩}

The child symbol is formatted (where appropriate) according to:

---

---

This just does \glslongextraSymbolFmt{⟨label⟩} by default.

The styles that have a location list column format the list for child entries using:

---

---

This just does ⟨locations⟩ and ignores the level and label.

The letter group headings are formatted according to:

---

---

which does nothing by default. The first argument ⟨n⟩ is the number of columns in the table. The second argument ⟨label⟩ is the group label (not the title, although they may happen to be the same).

This can be redefined to show the group heading. For example:

\renewcommand*{\glslongextraGroupHeading}[2]{%
  \glsxtrgetgrouptitle{#2}{\thisgrptitle}%
  \glslongextraHeaderFmt{\thisgrptitle}%
  \tabularnewline
  \noalign{\vskip\normalbaselineskip}%
}

This ignores the first argument and just puts the group title in the first column formatted according to \glslongextraHeaderFmt (to match the column header).

Remember that you can also adjust the styles through category attributes. The name column’s title is given by \entryname, the description column’s title is given by \descriptionname and (where present) the symbol column’s title is given by \symbolname, as for the other long styles that have headers.

long-name-desc This is like the longragged-booktabs style but doesn’t show the location list (regardless of the nonumberlist option). The name is shown in the first column and the description in the second.

The symbol is not displayed. The header row is produced with:

---

\glslongextraNameDescHeader

---

This essentially uses the same code as for longragged-booktabs but makes it easier to adjust the header without having to define a new style. This is defined as:

     \glslongextraNameDescTabularHeader\endhead
     \glslongextraNameDescTabularFooter\endfoot

where:

---

\glslongextraNameDescTabularHeader

---

sets up the header and

---

\glslongextraNameDescTabularFooter

---

sets up the footer. If you have setup the tabular version of this style then the above two commands are used at the start and end of the tabular environment (and \glslongextraNameDescHeader isn’t used).

For example, to simply remove the header and footer (for the default longtable version of the style):

     \renewcommand{\glslongextraNameDescHeader}{}

Or to change the name alignment to centred:

     \renewcommand{\glslongextraNameAlign}{c}

long-name-desc-loc This is like the long-name-desc style but has a third column for the location list. The longtable header is given by:

---

\glslongextraNameDescLocationHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraNameDescLocationTabularHeader

---

and

---

\glslongextraNameDescLocationTabularFooter

---

long-desc-name This is like the long-name-desc style but swaps the columns. Note that if the entry counter is displayed it will appear at the start of the second column by default. The longtable header is formatted according to:

---

\glslongextraDescNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraDescNameTabularHeader

---

and

---

\glslongextraDescNameTabularFooter

---

long-loc-desc-name This has three columns containing the location list, description and name. The longtable header is formatted according to:

---

\glslongextraLocationDescNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraLocationDescNameTabularHeader

---

and

---

\glslongextraLocationDescNameTabularFooter

---

long-name-desc-sym This is has three columns, with the name in the first, the description in the second and the symbol in the third.

The longtable header row is produced with:

---

\glslongextraNameDescSymHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraNameDescSymTabularHeader

---

and

---

\glslongextraNameDescSymTabularFooter

---

long-name-desc-sym-loc This is has four columns, with the name in the first, the description in the second, the symbol in the third and the location list in the fourth.

The longtable header row is produced with:

---

\glslongextraNameDescSymLocationHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraNameDescSymLocationTabularHeader

---

and

---

\glslongextraNameDescSymLocationTabularFooter

---

long-name-sym-desc This is like the long-name-desc-sym but the second and third column are swapped. The longtable header row is given by:

---

\glslongextraNameSymDescHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraNameSymDescTabularHeader

---

and

---

\glslongextraNameSymDescTabularFooter

---

long-name-sym-desc-loc This is like the long-name-desc-sym-loc but the second and third column are swapped. The longtable header row is given by:

---

\glslongextraNameSymDescLocationHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraNameSymDescLocationTabularHeader

---

and

---

\glslongextraNameSymDescLocationTabularFooter

---

long-sym-desc-name This has the symbol in the first column, the description in the second and the name in the third. The longtable header row is given by:

---

\glslongextraSymDescNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraSymDescNameTabularHeader

---

and

---

\glslongextraSymDescNameTabularFooter

---

long-loc-sym-desc-name This has the location list in the first column, symbol in the second column, the description in the third and the name in the fourth. The longtable header row is given by:

---

\glslongextraLocationSymDescNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraLocationSymDescNameTabularHeader

---

and

---

\glslongextraLocationSymDescNameTabularFooter

---

long-desc-sym-name This has the description in the first column, the symbol in the second and the name in the third. The longtable header row is given by:

---

\glslongextraDescSymNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraDescSymNameTabularHeader

---

and

---

\glslongextraDescSymNameTabularFooter

---

long-loc-desc-sym-name This has the location list in the first column, the description in the second column, the symbol in the third and the name in the fourth. The longtable header row is given by:

---

\glslongextraLocationDescSymNameHeader

---

which similarly defined in terms of the commands used for the tabular version:

---

\glslongextraLocationDescSymNameTabularHeader

---

and

---

\glslongextraLocationDescSymNameTabularFooter

---

Top

3.3 glossary-topic package

As from version 1.40, the glossaries-extra package comes with the supplementary package glossary-topic that provides glossary styles designed for hierarchical glossaries where the top-level entries are topic titles. This package automatically loads the multicols package. If the glossary-tree package is also loaded then commands like \glssetwidest can be used on these styles in much the same way as for the alttree style. If a widest value isn’t set then these styles behave more like the tree style.

topic

This style is similar to the tree style but the indentation doesn’t start until the second sub-item level. The top-level entries have the name displayed in a larger font with the description following in a new paragraph. This style doesn’t support the nogroupskip setting.

topicmcols

This style is like the topic style but the sub-entries are placed inside a multicols environment. The environment name is supplied in the value of the command:

---

\glstopicColsEnv

---

You can change this to the starred form. For example:

     \renewcommand{\glstopicColsEnv}{multicols*}

The number of columns is given by the command:

---

\glstopicCols

---

The default value is 2.

Both styles use the following commands.

---

---

This command is a length that’s used for the paragraph indentation in any multi-paragraph description for top-level entries, but not for the first paragraph (at the start of the description) which isn’t indented.

---

---

This command is a length that’s used to calculate the hanging indentation for sub-entries. The level 1 sub-entries don’t indent the name. Level n sub-entries have the name indented by (n- 1)×\glstopicSubIndent. The hanging indent depends on whether or not a widest name has been set for the level.

As from v1.46, there is also a length for additional indentation used in the second paragraph onwards for child entries with multi-paragraph descriptions:

---

---

This is initialised to \parindent when glossary-topic is loaded.

---

---

Hook used at the start of the glossary. Does nothing by default.

Although the styles don’t support letter groups by default, if you have many topics (top-level entries) and you feel that it would help the reader to divide them up into headed letter groups, you can redefine:

---

---

This does nothing by default. If you want to redefine it, you can fetch the title corresponding to the group label with \glsxtrgetgrouptitle. For example:

\renewcommand*{\glstopicGroupHeading}[1]{%
  \glsxtrgetgrouptitle{#1}{\thisgrptitle}%
  \section*{\thisgrptitle}%
}

Remember that if you use bib2gls, you will need the --group (or -g) switch to support this.

---

---

Used to format the name, symbol, description and location list for the top-level entries. This starts with a paragraph break followed by:

---

---

which defaults to \medskip. There is then a hook:

---

---

which does nothing by default, but may be redefined. For example, to add a line to the table of contents. The name and symbol are set in the form of a title using:

---

---

This uses \Glossentryname which converts the first letter to upper case. If there’s a symbol, this is added in parentheses. Both name and symbol (if present) are encapsulated by

---

---

This uses a bold, large font by default.

If the entry has the description key set (tested with \ifglshasdesc) then a paragraph break is inserted followed by:

---

---

which defaults to \smallskip. This is followed by the description which is formatted according to:

---

---

This just does \Glossentrydesc{⟨label⟩} followed by the post-description hook.

A paragraph break followed by:

---

---

comes next regardless of whether or not the description was displayed. This defaults to \smallskip. This is then followed by:

---

---

which may be used to display the location list, but does nothing by default.

The sub-entries first set up the paragraph and hanging indentations using:

---

---

This uses:

---

---

to determine if a widest name has been set for the given level.

The sub-entry has its information displayed using:

---

---

This encapsulates the name with:

---

---

By default this just uses \textbf. This is followed by:

---

---

which defaults to \quad. The name and separator are passed in the ⟨text⟩ argument of:

---

---

If a widest name was set for the given level, this will put ⟨text⟩ inside a box of that width otherwise it just does ⟨text⟩.

This is followed by the symbol in parentheses if set. Then, if the description is set, the description and post-description hook are displayed followed by:

---

---

(This command isn’t used if the description isn’t set.)

Finally the location list is displayed using:

---

---

which just does ⟨location⟩ by default.

4. Abbreviations

The new abbreviation system provided by glossaries-extra is more flexible than the acronym handling provided by the base glossaries package. The glossaries-extra package modifies the underlying formatting used by \gls (and its variants) so that terms that are recognised as abbreviations can have their formatting dealt with by the style. You therefore need to select an abbreviation style that ensures that \gls (and its variants) displays the desired output. See the file sample-abbr-styles.pdf for samples of all provided abbreviation styles.

This lack of flexibility in \glsfirst can be demonstrated with the following document:

\documentclass{article}
\usepackage{glossaries-extra}
\setabbreviationstyle{footnote}
\newabbreviation{ex}{EX}{Example}
\begin{document}
Compare \gls{ex}['s] with \glsfirst{ex}['s].
\end{document}

The first use \gls has the footnote marker after the inserted material “EX’s¹” but \glsfirst has it before the inserted material “EX¹’s” which is inappropriate. Note that if the style is changed to postfootnote, the footnote marker appears after the inserted material as \footnote is in the post-link hook.

There are some instances where \glstext can be useful. It’s used internally by \glsfmttext, which in turn is used by \glsseelistformat if the entry has a short form. If \glsfmtshort is used, this enforces the short form, but if \glsfmttext is used, then the long form will be used for the “noshort” styles, which is more appropriate. In this particular situation, there’s no need to worry about inserted material as the final optional argument isn’t supported by \glsfmttext.

Abbreviations include acronyms (words formed from initial letters, such as “laser”), initialisms (initial letters of a phrase, such as “html”, that aren’t pronounced as words) and contractions (where parts of words are omitted, often replaced by an apostrophe, such as “don’t”). The “acronym” code provided by the glossaries package is misnamed as it’s more often than not used for initialisms instead. Acronyms tend not to be expanded on first use (although they may need to be described for readers unfamiliar with the term). They are therefore more like a regular term, which may or may not require a description in the glossary.

The glossaries-extra package corrects this misnomer, and provides better abbreviation handling, with

---

---

This sets the category key to abbreviation by default, but that value may be overridden in ⟨options⟩. The category may have attributes that modify the way abbreviations are defined. For example, the insertdots attribute will automatically insert full stops (periods) into ⟨short⟩ or the noshortplural attribute will set the default value of the shortplural key to just ⟨short⟩ (without appending the plural suffix). See §7 Categories for further details.

See §2.8 Nested Links regarding the pitfalls of using commands like \gls or \glsxtrshort within ⟨short⟩ or ⟨long⟩.

The \newacronym command provided by the glossaries package is redefined by glossaries-extra to use \newabbreviation with the category set to acronym (see also §2.9 Acronym Style Modifications) so

---

---

is now equivalent to
\newabbreviation[type=\acronymtype,category=acronym,⟨options⟩] {⟨label⟩} {⟨short⟩} {⟨long⟩}

The \newabbreviation command is superficially similar to the glossaries package’s \newacronym but you can apply different styles to different categories. The default style is short-nolong for entries in the acronym category and short-long for entries in the abbreviation category. (These aren’t the same as the acronym styles provided by the glossaries package, although they may produce similar results.)

The way the abbreviations are displayed by commands like \gls varies according to the abbreviation style. The styles are set according to the entry’s category so, unlike the base glossaries package, you can have different abbreviation styles within the same glossary.

There are two types of full forms. The display full form, which is used on first use by commands like \gls and the inline full form, which is used by commands like \glsxtrfull. For some of the abbreviation styles, such as long-short, the display and inline forms are the same. In the case of styles such as short-nolong or short-footnote, the display and inline full forms are different.

These formatting commands aren’t stored in the short, shortplural, long or longplural fields, which means they won’t be used within commands like \glsentryshort (but they are used within commands like \glsxtrshort and \glsfmtshort). Note that \glsxtrlong and the case-changing variants don’t use \glsfirstlongfont.

You can apply the formatting command used for the short form to some arbitrary text using

---

---

where ⟨category⟩ is the category label that identifies the abbreviation style. Similarly for the formatting command use by the long form:

---

---

You can’t use the acronym commands provided by the base package with the new abbreviations provided by the glossaries-extra package. The style commands that replace \setacronymstyle, \acrshort etc are described in §4.2 Abbreviation Styles. The acronymlists package option and associated commands aren’t supported. The \forallacronyms command, which iterates over all acronym lists, should be replaced with:

---

---

Top

4.1 Tagging Initials

If you would like to tag the initial letters in the long form such that those letters are underlined in the glossary but not in the main part of the document, you can use

---

---

before you define your abbreviations.

This command (robustly) defines ⟨cs⟩ (a control sequence) to accept a single argument, which is the letter (or letters) that needs to be tagged. The normal behaviour of this command within the document is to simply do its argument, but in the glossary it’s activated for those categories that have the tagging attribute set to “true”. For those cases it will use

---

---

This command defaults to \underline{⟨text⟩} but may be redefined as required.

The control sequence ⟨cs⟩ can’t already be defined when used with the unstarred version of \GlsXtrEnableInitialTagging for safety reasons. The starred version will overwrite any previous definition of ⟨cs⟩. As with redefining any commands, ensure that you don’t redefine something important. In fact, just forget the existence of the starred version and let’s pretend I didn’t mention it.

The first argument of \GlsXtrEnableInitialTagging is a comma-separated list of category names. The tagging attribute will automatically be set for those categories. You can later set this attribute for other categories (see §7 Categories) but this must be done before the glossary is displayed.

The accompanying sample file sample-mixtures.tex uses initial tagging for both the acronym and abbreviation categories:

\GlsXtrEnableInitialTagging{acronym,abbreviation}{\itag}

This defines the command \itag which can be used in the definitions. For example:

\newacronym
 [description={a system for detecting the location and
 speed of ships, aircraft, etc, through the use of radio
 waves}% description of this term
 ]
 {radar}% identifying label
 {radar}% short form (i.e. the word)
 {\itag{ra}dio \itag{d}etection \itag{a}nd \itag{r}anging}
\newabbreviation{xml}{XML}
 {e\itag{x}tensible \itag{m}arkup \itag{l}anguage}

The underlining of the tagged letters only occurs in the glossary and then only for entries with the tagging attribute set.

Top

4.2 Abbreviation Styles

The abbreviation style must be set before abbreviations are defined using:

---

---

where ⟨style-name⟩ is the name of the style and ⟨category⟩ is the category label (abbreviation by default). New abbreviations will pick up the current style according to their given category. If there is no style set for the category, the fallback is the style for the abbreviation category. Some styles may automatically modify one or more of the attributes associated with the given category. For example, the long-noshort and short-nolong styles set the regular attribute to true.

Note that \setacronymstyle is disabled by glossaries-extra. Use

instead. The original acronym interface can be restored with \RestoreAcronyms (see §2.9 Acronym Style Modifications). However the original acronym interface is incompatible with all the commands described here.

Abbreviations can be used with the standard glossaries commands, such as \gls, but don’t use the acronym commands like \acrshort (which use \acronymfont). The short form can be produced with:

---

---

(Use this instead of \acrshort.)

The long form can be produced with

---

---

(Use this instead of \acrlong.)

The inline full form can be produced with

---

---

(This this instead of \acrfull.)

As mentioned earlier, the inline full form may not necessarily match the format used on first use with \gls. For example, the short-nolong style only displays the short form on first use, but the full form will display the long form followed by the short form in parentheses.

The arguments ⟨options⟩, ⟨label⟩ and ⟨insert⟩ are the same as for commands such as \glstext. There are also analogous case-changing commands:

First letter upper case short form:

---

---

First letter upper case long form:

---

---

First letter upper case inline full form:

---

---

All upper case short form:

---

---

All upper case long form:

---

---

All upper case inline full form:

---

---

Plural forms are also available.

Short form plurals:

---

---

---

---

---

---

Long form plurals:

---

---

---

---

---

---

Full form plurals:

---

---

---

---

---

---

Top

4.3 Shortcut Commands

The abbreviation shortcut commands can be enabled using the shortcuts=abbreviations package option (or shortcuts=abbr) or shortcuts=ac. (You can use both settings at the same time.) The provided shortcut commands listed in table 4.1.

Top

4.4 Predefined Abbreviation Styles

There are two types of abbreviation styles: those that treat the abbreviation as a regular entry (so that \gls uses \glsgenentryfmt) and those that don’t treat the abbreviation as a regular entry (so that \gls uses \glsxtrgenabbrvfmt).

The regular entry abbreviation styles set the regular attribute to “true” for the category assigned to each abbreviation with that style. This means that on first use, \gls uses the value of the first field and on subsequent use \gls uses the value of the text field (and analogously for the plural and case-changing versions). The short and long fields are set as appropriate and may be accessed through commands like \glsxtrshort.

The other abbreviation styles don’t modify the regular attribute. The first and text fields (and their plural forms) are set and can be accessed through commands like \glsfirst, but they aren’t used by commands like \gls, which instead use the short form (stored in the short key) and the display full format (through commands like \glsxtrfullformat that are defined by the style).

In both cases, the first use of \gls may not match the text produced by \glsfirst (and likewise for the plural and case-changing versions).

The sample file sample-abbr-styles.tex demonstrates all predefined styles described here.

Many of the styles have helper commands in the form \glsxtr...name and \glsxtr...sort that are expanded within the name and sort fields when the abbreviation is defined. These commands may use the helper token registers available within \newabbreviation, such as \glsshorttok. The contents of a register can be accessed with \the⟨register⟩. It’s essential that these expand when the abbreviation is defined so don’t hide them behind no-expandable content if you redefine these helper commands.

The parenthetical styles, such as long-short, use

---

---

to set the parenthetical material. This just puts parentheses around the text: (⟨text⟩).

The basic abbreviation styles, such as long-short and short-long use

---

---

for the short form. This just does ⟨text⟩ by default. (That is, no font change is applied.) On first use,

---

---

is used instead. By default, this just does \glsabbrvdefaultfont. The long form is formatted according to

---

---

which again just does ⟨text⟩ (no font change). On first use,

---

---

is used instead. This just does \glslongdefaultfont. The plural suffix used for the short form is given by

---

---

which defaults to \glspluralsuffix.

The small-cap styles, such as long-short-sc and short-sc-long, use

---

---

which uses \textsc.^4.1 On first use

---

---

is used instead. This uses \glsabbrvscfont by default. So redefine, \glsabbrvscfont to change first and subsequent uses or \glsfirstabbrvscfont to change just the first use.

The long form for the small-cap styles uses \glslongdefaultfont or \glsfirstlongdefaultfont, as with the basic style. The suffix is given by

---

---

This is defined as

\newcommand*{\glsxtrscsuffix}{\glstextup{\glsxtrabbrvpluralsuffix}}

The \glstextup command is provided by glossaries and is used to switch off the small caps font for the suffix. If you override the default short plural using the shortplural key when you define the abbreviation you will need to make the appropriate adjustment if necessary. (Remember that the default plural suffix behaviour can be modified through the use of the aposplural and noshortplural attributes. See §7 Categories for further details.)

The small styles, such as long-short-sm and short-sm-long, use

---

---

which uses \textsmaller. (This requires the relsizes package, which isn’t loaded by glossaries-extra, so must be loaded explicitly.) On first use

---

---

is used instead. This uses \glsabbrvsmfont by default.

The long form for the smaller styles uses \glslongdefaultfont or \glsfirstlongdefaultfont, as with the basic style. The suffix is given by

---

---

which defaults to just \glsxtrabbrvpluralsuffix.

The “short-em” (emphasize short) styles, such as long-short-em or short-em-long, use

---

---

On first use

---

---

is used instead. This uses \glsabbrvemfont by default. The suffix is given by

---

---

which defaults to just \glsxtrabbrvpluralsuffix. The long form is as for the basic style unless the style is a “long-em” style.

The “long-em” (emphasize long) styles, such as long-em-short-em or short-em-long-em, use

---

---

instead of \glsfirstlongdefaultfont{⟨long-form⟩} and

---

---

instead of \glslongdefaultfont{⟨long-form⟩}. The first form \glsfirstlongemfont is initialised to use \glslongemfont.

The user styles have similar commands:

---

---

for the short form,

---

---

for the first use short form,

---

---

for the long form,

---

---

for the first use long form, and

---

---

for the short plural suffix.

Similarly for the hyphen styles:

---

---

for the short form,

---

---

for the first use short form,

---

---

for the long form,

---

---

for the first use long form, and

---

---

for the short plural suffix.

Similarly for the “only” styles, such as long-only-short-only:

---

---

for the short form,

---

---

for the first use short form,

---

---

for the long form,

---

---

for the first use long form, and

---

---

for the short plural suffix.

The “short-sc-only” styles, such as long-only-short-sc-only:

---

---

for the short form (which uses \glsabbrvscfont), and

---

---

which uses \glsabbrvsconlyfont. The short plural suffix is obtained from:

---

---

This uses \glsxtrscsuffix. For other font types, use the one of “short-only” styles and redefine \glsabbrvonlyfont as required. (The smallcaps style is a special case as its plural suffix is awkward.)

The “postshort-sc-user” styles, such as long-postshort-sc-user:

---

---

for the short form (which uses \glsabbrvscfont), and

---

---

which uses \glsabbrvscuserfont. The short plural suffix is obtained from:

---

---

This uses \glsxtrscsuffix.

Note that by default inserted material (provided in the final optional argument of commands like \gls), is placed outside the font command in the predefined styles. To move it inside, use:

---

---

This applies to all the predefined styles. For example:

\setabbreviationstyle{long-short}
\renewcommand*{\glsfirstlongdefaultfont}[1]{\emph{#1}}
\glsxtrinsertinsidetrue

This will make the long form and the inserted text emphasized, whereas the default (without \glsxtrinsertinsidetrue) would place the inserted text outside of the emphasized font.

Note that for some styles, such as the short-long, the inserted text would be placed inside the font command for the short form (rather than the long form in the above example).

Remember that \textsc renders lowercase letters as small capitals. Uppercase letters are rendered as normal uppercase letters, so if you specify the short form in uppercase, you won’t get small capitals unless you redefine \glsabbrvscfont to convert its argument to lowercase. For example:

\renewcommand*{\glsabbrvscfont}[1]{\textsc{\MakeLowercase{#1}}}

If you want to easily switch between the “sc” and “sm” styles, you may find it easier to redefine this command to convert case:

\renewcommand*{\glsabbrvscfont}[1]{\textsc{\MakeTextLowercase{#1}}}
\renewcommand*{\glsabbrvsmfont}[1]{\textsmaller{\MakeTextUppercase{#1}}}

Some of the styles use

---

---

as a separator between the long and short forms. This is defined as a space by default, but may be changed as required. For example:

\renewcommand*{\glsxtrfullsep}[1]{~}

or

\renewcommand*{\glsxtrfullsep}[1]{\glsacspace{#1}}

The new naming scheme for abbreviation styles is as follows:

• ⟨field1⟩[-⟨modifier1⟩]-[post]⟨field2⟩[-⟨modifier2⟩][-user]

  This is for the parenthetical styles. The -⟨modifier⟩ parts may be omitted. These styles display ⟨field1⟩ followed by ⟨field2⟩ in parentheses. If ⟨field1⟩ or ⟨field2⟩ starts with “no” then that element is omitted from the display style (no parenthetical part) but is included in the inline style.

  If the -⟨modifier⟩ part is present, then the field has a font changing command applied to it. The special modifier -only indicates that field is only present according to whether or not the entry has been used.

  If post is present then ⟨field2⟩ is placed after the link-text using the post-link hook.

  If the -user part is present, then the user1 value, if provided, is inserted into the parenthetical material . (The field used for the inserted material may be changed.)

  Examples:

  • long-noshort-sc: ⟨field1⟩ is the long form, the short form is set in smallcaps but omitted in the display style.
  • long-em-short-em: both the long form and the short form are emphasized. The short form is in parentheses.
  • long-short-em: the short form is emphasized but not the long form. The short form is in parentheses.
  • long-short-user: if the user1 key has been set, this produces the style ⟨long⟩ (⟨short⟩, ⟨user1⟩) otherwise it just produces ⟨long⟩ (⟨short⟩).
  • long-hyphen-postshort-hyphen: the short form and the inserted material (provided by the final optional argument of commands like \gls) is moved to the post-link hook. The long form is formatted according to \glslonghyphenfont (or \glsfirstlonghyphenfont on first use). The short form is formatted according to \glsabbrvhyphenfont (or \glsfirstabbrvhyphenfont on first use).
• ⟨style⟩-noreg

  Some styles set the regular attribute. In some cases, there’s a version of the style that doesn’t set this attribute. For example, long-em-noshort-em sets the regular attribute. The long-em-noshort-em-noreg style is a minor variation that style that doesn’t set the attribute.

  There are a few “noshort” styles, such as long-hyphen-noshort-noreg, that have “-noreg” version without a regular version. This is because the style won’t work properly with the regular set, but the naming scheme is maintained for consistency with the other “noshort” styles.

• ⟨field1⟩[-⟨modifier1⟩]-[post]footnote

  The display style uses ⟨field1⟩ followed by a footnote with the other field in it. If post is present then the footnote is placed after the link-text using the post-link hook. The inline style does ⟨field1⟩ followed by the other field in parentheses.

  If -⟨modifier1⟩ is present, ⟨field1⟩ has a font-changing command applied to it.

  Examples:

  • short-footnote: short form in the text with the long form in the footnote.
  • short-sc-postfootnote: short form in smallcaps with the long form in the footnote outside of the link-text.
  Take care with the footnote styles. Remember that there are some situations where \footnote doesn’t work.

• ⟨style⟩-desc

  Like ⟨style⟩ but the description key must be provided when defining abbreviations with this style.

  Examples:

  • short-long-desc: like short-long but requires a description.
  • short-em-footnote-desc: like short-em-footnote but requires a description.

Not all combinations that fit the above syntax are provided. Pre-version 1.04 styles that didn’t fit this naming scheme are either provided with a synonym (where the former name wasn’t ambiguous) or provided with a deprecated synonym (where the former name was confusing). The deprecated style names generate a warning using:

---

---

where ⟨old-name⟩ is the deprecated name and ⟨new-name⟩ is the preferred name. You can suppress these warnings by redefining this command to do nothing.

Top

4.4.1 Predefined Abbreviation Styles that Set the Regular Attribute

The following abbreviation styles set the regular attribute to “true” for all categories that have abbreviations defined with any of these styles.

short-nolong

This only displays the short form on first use. The name is set to the short form through the command

---

\glsxtrshortnolongname

---

(Similarly for the other short⟨modifier⟩-nolong⟨modifier⟩ styles, unless indicated otherwise.) This command is expanded as the entry is defined, so any redefinition must be done before \newabbreviation (or \newacronym) for it to take effect. Make sure to \protect any formatting commands (or anything else that shouldn’t be expanded).

The description is set to the long form. The inline full form displays ⟨short⟩ (⟨long⟩). The long form on its own can be displayed through commands like \glsxtrlong.

short

A synonym for short-nolong.

nolong-short

Like short-nolong but the inline full form displays ⟨long⟩ (⟨short⟩).

short-sc-nolong

Like short-nolong but redefines \glsabbrvfont to use \glsabbrvscfont (which defaults to \textsc).

short-sc

A synonym for short-sc-nolong

nolong-short-sc

Like short-sc-nolong but the inline full form displays ⟨long⟩ (⟨short⟩). The name is still obtained from \glsxtrshortnolongname (similarly for the other styles in the form nolong⟨modifier⟩-short⟨modifier⟩ unless indicated otherwise).

short-sm-nolong

Like short-nolong but redefines \glsabbrvfont to use \glsabbrvsmfont (which defaults to \textsmaller).

short-sm

A synonym for short-sm-nolong.

nolong-short-sm

Like short-sm-nolong but the inline full form displays ⟨long⟩ (⟨short⟩).

short-em-nolong

Like short-nolong but redefines \glsabbrvfont to use \glsabbrvemfont.

short-em

A synonym for short-em-nolong

nolong-short-em

Like short-em-nolong but the inline full form displays ⟨long⟩ (⟨short⟩).

short-nolong-desc

Like the short-nolong style, but the name is set to the full form obtained by expanding

---

\glsxtrshortdescname

---

(Similarly for the other short⟨modifier⟩-nolong⟨modifier⟩-desc styles, unless indicated otherwise.) This command is expanded when the entry is defined, so \protect fragile and formatting commands and only redefine this command before \newabbreviation (or \newacronym).

The description must be supplied by the user. You may prefer to use the short-nolong style with the post-description hook set to display the long form and override the description key. (See the sample file sample-acronym-desc.tex.)

short-desc

A synonym for short-nolong-desc.

short-sc-nolong-desc

Like short-nolong but redefines \glsabbrvfont to use \glsabbrvscfont (which defaults to \textsc).

short-sc-desc

A synonym for short-sc-nolong-desc.

short-sm-nolong-desc

Like short-nolong-desc but redefines \glsabbrvfont to use \glsabbrvsmfont (which defaults to \textsmaller).

short-sm-desc

A synonym for short-sm-nolong-desc.

short-em-nolong-desc

Like short-nolong-desc but redefines \glsabbrvfont to use \glsabbrvemfont.

short-em-desc

A synonym for short-em-nolong-desc.

long-noshort-desc

This style only displays the long form, regardless of first or subsequent use of commands \gls. The short form may be accessed through commands like \glsxtrshort. The inline full form displays ⟨long⟩ (⟨short⟩).

The sort key are set to the long form. The name key is also set to the long form, but this is done by expanding

---

\glsxtrlongnoshortdescname

---

(Similarly for the other long⟨modifier⟩-noshort⟨modifier⟩-desc styles, unless indicated otherwise.) This command should only be redefined before abbreviations are defined, and any fragile or formatting commands within it need protecting.

The description must be provided by the user. The predefined glossary styles won’t display the short form. You can use the post-description hook to automatically append the short form to the description. The inline full form will display ⟨long⟩ (⟨short⟩).

long-desc

A synonym for long-noshort-desc.

long-noshort-sc-desc

Like the long-noshort-desc style but the short form (accessed through commands like \glsxtrshort) use \glsabbrvscfont. (This style was originally called long-desc-sc. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-noshort-sm-desc

Like long-noshort-desc but redefines \glsabbrvfont to use \glsabbrvsmfont. (This style was originally called long-desc-sm. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-noshort-em-desc

Like long-noshort-desc but redefines \glsabbrvfont to use \glsabbrvemfont. The long form isn’t emphasized. (This style was originally called long-desc-em. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-em-noshort-em-desc

New to version 1.04, like long-noshort-desc but redefines \glsabbrvfont to use \glsabbrvemfont. The long form uses \glsfirstlongemfont and \glslongemfont.

long-noshort

This style doesn’t really make sense if you don’t use the short form anywhere in the document, but is provided for completeness. This is like the long-noshort-desc style, but the sort key is set to the short form. The name key is also set to the short form, but this is done by expanding

---

\glsxtrlongnoshortname

---

(Similarly for other long⟨modifier⟩-noshort⟨modifier⟩ styles, unless indicated otherwise.) This command should only be redefined before abbreviations are defined, and fragile or formatting commands should be protected.

The description is set to the long form.

long

A synonym for long-noshort

long-noshort-sc

Like the long-noshort style but the short form (accessed through commands like \glsxtrshort) use \glsabbrvscfont. (This style was originally called long-sc. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-noshort-sm

Like long-noshort but redefines \glsabbrvfont to use \glsabbrvsmfont. (This style was originally called long-sm. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-noshort-em

This style is like long-noshort but redefines \glsabbrvfont to use \glsabbrvemfont. The long form isn’t emphasized. (This style was originally called long-em. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

long-em-noshort-em

New to version 1.04, this style is like long-noshort but redefines \glsabbrvfont to use \glsabbrvemfont, \glsfirstlongfont to use \glsfirstlongemfont and \glslongfont to use \glslongemfont. The short form isn’t used by commands like \gls, but can be obtained using \glsxtrshort. The related style long-em-noshort-em-noreg doesn’t set the regular attribute.

Top

4.4.2 Predefined Abbreviation Styles that Don’t Set the Regular Attribute

The following abbreviation styles will set the regular attribute to “false” if it has previously been set. If it hasn’t already been set, it’s left unset. Other attributes may also be set, depending on the style.

long-short

On first use, this style uses the format ⟨long⟩ (⟨short⟩). The inline and display full forms are the same. The sort key is set to the short form. The name is also set to the short form through

---

\glsxtrlongshortname

---

(Similarly for other long⟨modifier⟩-short⟨modifier⟩ styles, unless indicated otherwise.) Any redefinition of this command must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

The description is set to the long form. The long and short forms are separated by \glsxtrfullsep. If you want to insert material within the parentheses (such as a translation), try the long-short-user style.

long-short-sc

Like long-short but redefines \glsabbrvfont to use \glsabbrvscfont.

long-short-sm

Like long-short but redefines \glsabbrvfont to use \glsabbrvsmfont.

long-short-em

Like long-short but redefines \glsabbrvfont to use \glsabbrvemfont.

long-em-short-em

New to version 1.04, this style is like long-short-em but redefines \glsfirstlongfont to use \glsfirstlongemfont.

long-only-short-only

New to version 1.17, this style only shows the long form on first use and only shows the short form on subsequent use. The inline full form \glsxtrfull shows the long form followed by the short form in parentheses. The name field is obtained from

---

\glsxtronlyname

---

Any redefinition of this command must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

long-only-short-sc-only

New to version 1.48, this is like long-only-short-only but uses smallcaps for the short form. The name is obtained from:

---

\glsxtrsconlyname

---

long-only-short-only-desc

New to version 1.17, this style is like long-only-short-only but the user must supply the description. The name field is obtained from

---

\glsxtronlydescname

---

The sort value is obtained from:

---

\glsxtronlydescsort

---

Any redefinition of these commands must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

long-only-short-sc-only-desc

New to version 1.48, this style is like long-only-short-only-desc but it uses smallcaps for the short form. The name is obtained from:

---

\glsxtrsconlydescname

---

This defaults to \glsxtronlydescname.

The sort value is obtained from:

---

\glsxtrsconlydescsort

---

This defaults to \glsxtronlydescsort.

long-em-noshort-em-noreg

New to version 1.17, this style is like long-em-noshort-em but doesn’t set the regular attribute.

long-short-user

This style was introduced in version 1.04. It’s like the long-short style but additional information can be inserted into the parenthetical material. This checks the value of the field given by

---

\glsxtruserfield

---

(which defaults to useri) using \ifglshasfield (provided by glossaries). If the field hasn’t been set, the style behaves like the long-short style and produces ⟨long⟩ (⟨short⟩) but if the field has been set, the contents of that field are inserted within the parentheses in the form ⟨long⟩ (⟨short⟩, ⟨field-value⟩). The format is governed by

---

\glsxtruserparen{⟨text⟩}{⟨label⟩}

---

where ⟨text⟩ is the short form (for the long-short-user style) or the long form (for the short-long-user style). This command first inserts a space using \glsxtrfullsep and then the parenthetical content (using \glsxtrparen). The description is set to

---

\glsuserdescription{⟨long⟩}{⟨label⟩}

---

The default definition ignores the ⟨label⟩ and encapsulates ⟨long⟩ with \glslonguserfont.

The name is obtained by expanding \glsxtrlongshortname (see above). The ⟨text⟩ argument includes the font formatting command, \glsfirstabbrvfont {⟨short⟩} in the case of the long-short-user style and \glsfirstlongfont{⟨long⟩} in the case of the short-long-user style.

For example:

     \setabbreviationstyle[acronym]{long-short-user}
     \newacronym{tug}{TUG}{\TeX\ User Group}
     \newacronym
      [user1={German Speaking \TeX\ User Group}]
      {dante}{DANTE}{Deutschsprachige Anwendervereinigung \TeX\ e.V}

On first use, \gls{tug} will appear as:

TeX User Group (TUG)

whereas \gls{dante} will appear as:

Deutschsprachige Anwendervereinigung TeX e.V (DANTE, German Speaking TeX User Group)

The short form is formatted according to

---

\glsabbrvuserfont{⟨text⟩}

---

and the plural suffix is given by

---

\glsxtrusersuffix

---

These may be redefined as appropriate. For example, if you want a smallcaps style, you can just set these commands to those used by the long-short-sc style:

     \renewcommand{\glsabbruserfont}[1]{\glsabbrvscfont{#1}}
     \renewcommand{\glsxtrusersuffix}{\glsxtrscsuffix}

long-noshort-noreg

This style is like long-noshort but it doesn’t set the regular attribute.

long-noshort-desc-noreg

This style is like long-noshort-desc but it doesn’t set the regular attribute.

long-short-desc

On first use, this style uses the format ⟨long⟩ (⟨short⟩). The inline and display full forms are the same. The name is set to the full form. The sort key is set to ⟨long⟩ (⟨short⟩). Before version 1.04, this was incorrectly set to the short form. If you want to revert back to this you can redefine

---

\glsxtrlongshortdescsort

---

For example:

     \renewcommand*{\glsxtrlongshortdescsort}{\the\glsshorttok}

The description must be supplied by the user. The long and short forms are separated by \glsxtrfullsep. The name field is obtained from

---

\glsxtrlongshortdescname

---

(Similarly for other long⟨modifier⟩-short⟨modifier⟩-desc styles, unless indicated otherwise.) Any redefinition of this command must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

long-short-sc-desc

Like long-short-desc but redefines \glsabbrvfont to use \glsabbrvscfont.

long-short-sm-desc

Like long-short-desc but redefines \glsabbrvfont to use \glsabbrvsmfont.

long-short-em-desc

Like long-short-desc but redefines \glsabbrvfont to use \glsabbrvemfont.

long-em-short-em-desc

New to version 1.04, this style is like long-short-em-desc but redefines \glsfirstlongfont to use \glsfirstlongemfont.

long-em-noshort-em-desc-noreg

New to version 1.17, this style is like long-em-noshort-em-desc but doesn’t set the regular attribute.

long-short-user-desc

New to version 1.04, this style is like a cross between the long-short-desc style and the long-short-user style. The display and inline forms are as for long-short-user and the name key is obtained from

---

\glsxtrlongshortuserdescname

---

Again, this should only be redefined before \newabbreviation (or \newacronym), and fragile and formatting commands need protecting.

The description key must be supplied in the optional argument of \newabbreviation (or \newacronym). The sort key is set to ⟨long⟩ (⟨short⟩) as per the long-short-desc style.

short-nolong-noreg

This is like short-nolong but doesn’t set the regular attribute.

nolong-short-noreg

This is like nolong-short but doesn’t set the regular attribute.

short-long

On first use, this style uses the format ⟨short⟩ (⟨long⟩). The inline and display full forms are the same. The name and sort keys are set to the short form. The description is set to the long form. The short and long forms are separated by \glsxtrfullsep. If you want to insert material within the parentheses (such as a translation), try the short-long-user style.

The name field is obtained from

---

\glsxtrshortlongname

---

(Similarly for other short⟨modifier⟩-long⟨modifier⟩ styles, unless indicated otherwise.) Any redefinition of this command must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

short-sc-long

Like short-long but redefines \glsabbrvfont to use \glsabbrvscfont.

short-sm-long

Like short-long but redefines \glsabbrvfont to use \glsabbrvsmfont.

short-em-long

Like short-long but redefines \glsabbrvfont to use \glsabbrvemfont.

short-em-long-em

New to version 1.04, this style is like short-em-long but redefines \glsfirstlongfont to use \glsfirstlongemfont.

short-long-user

New to version 1.04. This style is like the long-short-user style but with the long and short forms switched. The parenthetical material is governed by the same command \glsxtruserparen, but the first argument supplied to it is the long form instead of the short form. The name field is obtained by expanding

---

\glsxtrshortlongname

---

Again, this should only be redefined before \newabbreviation (or \newacronym) and commands that should be expanded need to be protected. The description is set to \glsuserdescription{⟨long⟩}{⟨label⟩}.

short-nolong-desc-noreg

This style is like short-nolong-desc but it doesn’t set the regular attribute.

short-long-desc

On first use, this style uses the format ⟨short⟩ (⟨long⟩). The inline and display full forms are the same. The name is set to the full form. The description must be supplied by the user. The short and long forms are separated by \glsxtrfullsep. The name field is obtained from

---

\glsxtrshortlongdescname

---

(Similarly for other short⟨modifier⟩-long⟨modifier⟩-desc styles, unless indicated otherwise.) Any redefinition of this command must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

short-sc-long-desc

Like short-long-desc but redefines \glsabbrvfont to use \glsabbrvscfont.

short-sm-long-desc

Like short-long-desc but redefines \glsabbrvfont to use \glsabbrvsmfont.

short-em-long-desc

Like short-long-desc but redefines \glsabbrvfont to use \glsabbrvemfont.

short-em-long-em-desc

New to version 1.04, this style is like short-em-long-desc but redefines \glsfirstlongfont to use \glsfirstlongemfont.

short-long-user-desc

New to version 1.04, this style is like a cross between the short-long-desc style and the short-long-user style. The display and inline forms are as for short-long-user, but the name key is obtained from

---

\glsxtrshortlonguserdescname

---

Again, this should only be redefined before \newabbreviation (or \newacronym), and fragile and formatting commands need protecting.

The description key must be supplied in the optional argument of \newabbreviation (or \newacronym).

short-footnote

On first use, this style displays the short form with the long form as a footnote. This style automatically sets the nohyperfirst attribute to “true” for the supplied category, so the first use won’t be hyperlinked (but the footnote marker may be, if the hyperref package is used).

The inline full form uses the ⟨short⟩ (⟨long⟩) style. The name is set to the short form. The description is set to the long form. The name key is obtained by expanding:

---

\glsxtrfootnotename

---

(Similarly for other short⟨modifier⟩-⟨modifier⟩footnote styles, unless indicated otherwise.) Again, this command should only be redefined before \newabbreviation (or \newacronym), and fragile or formatting commands should be protected from expansion.

As from version 1.05, all the footnote styles use:

---

\glsfirstlongfootnotefont{⟨text⟩}

---

to format the long form on first use or for the full form and

---

\glslongfootnotefont{⟨text⟩}

---

to format the long form elsewhere (for example, when used with \glsxtrlong).

As from version 1.07, all the footnote styles use:

---

\glsxtrabbrvfootnote{⟨label⟩}{⟨long⟩}

---

By default, this just does \footnote{⟨long⟩} (the first argument is ignored). For example, to make the footnote text link to the relevant place in the glossary:

     \renewcommand{\glsxtrabbrvfootnote}[2]{%
       \footnote{\glshyperlink[#2]{#1}}%
     }

or to include the short form with a hyperlink:

     \renewcommand{\glsxtrabbrvfootnote}[2]{%
       \footnote{\glshyperlink[\glsfmtshort{#1}]{#1}: #2}%
     }

Note that I haven’t used commands like \glsxtrshort to avoid interference (see §2.4 Entry Display Style Modifications and §2.8 Nested Links).

footnote

A synonym for short-footnote.

short-sc-footnote

Like short-footnote but redefines \glsabbrvfont to use \glsabbrvscfont. (This style was originally called footnote-sc. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-sc-footnote

Like short-footnote but redefines \glsabbrvfont to use \glsabbrvsmfont. (This style was originally called footnote-sm. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-em-footnote

Like short-footnote but redefines \glsabbrvfont to use \glsabbrvemfont. (This style was originally called footnote-em. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-footnote-desc

(New to v1.42.) This is similar to short-footnote, but the description must be supplied by the user. The name field is obtained from:

---

\glsxtrfootnotedescname

---

which defaults to ⟨short⟩ followed by ⟨long⟩ in parentheses, and the sort field is obtained from:

---

\glsxtrfootnotedescsort

---

which defaults to just the short form. (Similarly for other short⟨modifier⟩-[post]footnote-desc styles, unless indicated otherwise.)

Any redefinition of these commands must come before the abbreviations are defined as it expands on definition. Make sure you protect any commands that shouldn’t be expanded. The long form can be referenced with \the\glslongtok and the short form can be referenced with \the\glsshorttok.

footnote-desc

A synonym for short-footnote-desc.

short-sc-footnote-desc

Like short-footnote-desc but redefines \glsabbrvfont to use \glsabbrvscfont.

short-sm-footnote-desc

Like short-footnote-desc but redefines \glsabbrvfont to use \glsabbrvsmfont.

short-em-footnote-desc

Like short-footnote-desc but redefines \glsabbrvfont to use \glsabbrvemfont.

short-postfootnote

This is similar to the short-footnote style but doesn’t modify the category attribute. Instead it changes \glsxtrpostlink⟨category⟩ to insert the footnote after the link-text on first use. This will also defer the footnote until after any following punctuation character that’s recognised by \glsxtrifnextpunc.

The inline full form uses the ⟨short⟩ (⟨long⟩) style. The name is set to the short form. The description is set to the long form. Note that this style will change \glsxtrfull (and its variants) so that it fakes non-first use. (Otherwise the footnote would appear after the inline form.)

postfootnote

A synonym for short-postfootnote.

short-sc-postfootnote

Like short-postfootnote but redefines \glsabbrvfont to use \glsabbrvscfont. (This style was originally called postfootnote-sc. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-sm-postfootnote

Like short-postfootnote but redefines \glsabbrvfont to use \glsabbrvsmfont. (This style was originally called postfootnote-sm. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-em-postfootnote

Like short-postfootnote but redefines \glsabbrvfont to use \glsabbrvemfont. (This style was originally called postfootnote-em. Renamed in version 1.04, but original name retained as a deprecated synonym for backward-compatibility.)

short-postfootnote-desc

(New to v1.42.) This is similar to short-postfootnote, but the description must be supplied by the user. The name and sort are set as for short-footnote-desc.

postfootnote-desc

A synonym for short-postfootnote-desc.

short-sc-postfootnote-desc

Like short-postfootnote-desc but redefines \glsabbrvfont to use \glsabbrvscfont.

short-sm-postfootnote-desc

Like short-postfootnote-desc but redefines \glsabbrvfont to use \glsabbrvsmfont.

short-em-postfootnote-desc

Like short-postfootnote-desc but redefines \glsabbrvfont to use \glsabbrvemfont.

short-postlong-user

This style was introduced in version 1.12. It’s like the short-long-user style but defers the parenthetical material to after the link-text. This means that you don’t have such a long hyperlink (which can cause problems for the DVI LaTeX format) and it also means that the user supplied material can include a hyperlink to another location. The name key is obtained from \glsxtrshortlongname.

short-postlong-user-desc

This style was introduced in version 1.12. It’s like the above short-postlong-user style but the description must be specified. The name is obtained from \glsxtrshortlonguserdescname.

long-postshort-user

This style was introduced in version 1.12. It’s like the above short-postlong-user style but the long form is shown first and the short form is in the parenthetical material (as for long-short-user) style.

long-postshort-user-desc

This style was introduced in version 1.12. It’s like the above long-postshort-user style but the description must be specified. The name is obtained from \glsxtrlongshortuserdescname.

long-postshort-sc-user

This style (new to v1.48) is like long-postshort-user but uses smallcaps for the short form.

long-postshort-sc-user-desc

This style (new to v1.48) is like long-postshort-user-desc but uses smallcaps for the short form.

long-hyphen-short-hyphen

This style (new to v1.17) is designed to work with the markwords category attribute. The full form is formatted using

---

\glsxtrlonghyphenshort{⟨label⟩}{⟨long⟩}{⟨short⟩}{⟨insert⟩}

---

where ⟨insert⟩ is the inserted material provided in the final optional argument of commands like \insert. If ⟨insert⟩ start with a hyphen, then this locally redefines \glsxtrwordsep to a hyphen, which means that if the markwords attribute is set then the long form will become hyphenated. (If this attribute isn’t set, there’s no alteration to the way the long form is displayed.) The name key is obtained from \glsxtrlongshortname.

Unlike the other ⟨long⟩ (⟨short⟩) type of styles, such as long-short, this style also repeats the insertion in the parenthetical part, so that the first use form is:

\glsfirstlonghyphenfont{⟨long⟩}⟨insert⟩ (\glsfirstabbrvhyphenfont{⟨short⟩}⟨insert⟩)

The space before the parenthetical material is actually given by \glsxtrfullsep{⟨label⟩} which defaults to a space. The ⟨insert⟩ may be moved into the formatting commands according to the conditional \ifglsxtrinsertinside.

For example, if ip is defined using:

     \glssetcategoryattribute{english}{markwords}{true}
     \setabbreviationstyle[english]{long-hyphen-short-hyphen}
     \newabbreviation[category=english]{ip}{IP}{Internet Protocol}

then

     \gls{ip}[-Adressen]

will do

Internet-Protocol-Adressen (IP-Adressen)

on first use, whereas

     \gls{ip}[ Address]

will do

Internet Protocol Address (IP Address)

on first use.

Note that the hyphenation isn’t applied when using commands like \glsxtrlong. This means that

     \glsxtrlong{ip}[-Adressen]

will do

Internet Protocol-Adressen

If the markwords attribute hadn’t been set, then the first use of

     \gls{ip}[-Adressen]

would do

Internet Protocol-Adressen (IP-Adressen)

instead.

If the inserted material ⟨insert⟩ is likely to contain commands like \gls, then use long-hyphen-postshort-hyphen instead to avoid nested links.

If you want the short version in small-caps, you can just redefine \glsabbrvhyphenfont and \glsxtrhyphensuffix to use the small-caps versions. For example:

     \renewcommand{\glsabbrvhyphenfont}{\glsabbrvscfont}
     \renewcommand{\glsxtrhyphensuffix}{\glsxtrscsuffix}

Similarly for other font-changing variations.

long-hyphen-noshort-desc-noreg

New to version 1.17, this style is like long-hyphen-short-hyphen-desc except that the parenthetical part is omitted and the long form is displayed on subsequent use. The short form can be accessed with \glsxtrshort but just uses the default abbreviation font. There’s no regular version of this style as the regular form doesn’t have the flexibility to deal with the hyphen switch. The name is obtained from \glsxtrlongnoshortdescname.

long-hyphen-noshort-noreg

New to version 1.17, this style is like long-hyphen-noshort-desc-noreg but the name is set to the short form and the description is set to the long form.

long-hyphen-short-hyphen-desc

New to version 1.17. This is similar to long-hyphen-short-hyphen but the user supplies the description. The name is obtained from \glsxtrlongshortdescname.

long-hyphen-postshort-hyphen

New to version 1.17. This is similar to long-hyphen-short-hyphen but the inserted and parenthetical material are moved to the post-link hook. On first use, \gls{⟨label⟩}[⟨insert⟩] will do
\glsxtrlonghyphen{⟨long⟩}{⟨label⟩}{⟨insert⟩}\glsxtrposthyphenshort {⟨label⟩}⟨insert⟩
where

---

\glsxtrposthyphenshort{⟨label⟩}⟨insert⟩

---

is in the post-link hook. This uses the format:
⟨insert⟩ (\glsfirstabbrvhyphenfont{⟨short⟩}⟨isnert⟩)

The singular short form is always used here, even with \glspl. (Unlike long-hyphen-long-hyphen.)

The part in the link-text on first use:

---

\glsxtrlonghyphen{⟨long⟩}{⟨label⟩}{⟨insert⟩}

---

checks if ⟨insert⟩ starts with a hyphen. If it does, then \glsxtrwordsep is locally redefined to a hyphen. This command only uses ⟨insert⟩ to test if it starts with a hyphen. The actual insertion code isn’t typeset until the post-link hook and it’s also localised, which means that you can use commands like \gls in ⟨insert⟩ for this style without causing nested hyperlinks, but only for commands like \gls.

Don’t use \gls in the ⟨insert⟩ part for commands like \glsxtrfull, \glsxtrshort or \glsxtrlong.

The inline full display format used by commands like \glsxtrfull behaves differently to the first use of \gls with this style. It’s better to use \glsreset{⟨label⟩}\gls{⟨label⟩} if you want to ensure the full format.

long-hyphen-postshort-hyphen-desc

New to version 1.17. This is similar to long-hyphen-postshort-hyphen but the user supplies the description. The name is obtained from \glsxtrlongshortdescname.

short-hyphen-long-hyphen

This style (new to v1.17) is like long-hyphen-short-hyphen, except that the short form is displayed first followed by the long form in parentheses. The full form is formatted using

---

\glsxtrshorthyphenlong{⟨label⟩}{⟨short⟩}{⟨long⟩}{⟨insert⟩}

---

which behaves in an analogous way to \glsxtrlonghyphenshort. The name is obtained from \glsxtrshortlongname.

short-hyphen-long-hyphen-desc

New to version 1.17. This is similar to short-hyphen-long-hyphen but the user supplies the description. The name is obtained from \glsxtrshortlongdescname.

short-hyphen-postlong-hyphen

This style (new to v1.17) is like long-hyphen-postshort-hyphen, but the short form is displayed first followed by the long form in parentheses. On first use, \gls{⟨label⟩}[⟨insert⟩] will do
\glsxtrshorthyphen{⟨short⟩}{⟨label⟩}{⟨insert⟩}\glsxtrposthyphenlong {⟨label⟩}⟨insert⟩
where

---

\glsxtrposthyphenlong{⟨label⟩}⟨insert⟩

---

is in the post-link hook. These commands behave in an analogous manner to those used with long-hyphen-postshort-hyphen. The name is obtained from \glsxtrshortlongname.

Don’t use \gls in the ⟨insert⟩ part for commands like \glsxtrfull, \glsxtrshort or \glsxtrlong.

The inline full display format used by commands like \glsxtrfull behaves differently to the first use of \gls with this style. It’s better to use \glsreset{⟨label⟩}\gls{⟨label⟩} if you want to ensure the full format.

short-hyphen-postlong-hyphen-desc

New to version 1.17. This is similar to short-hyphen-postlong-hyphen but the user supplies the description. The name is obtained from \glsxtrshortlongdescname.

Top

4.5 Defining New Abbreviation Styles

New abbreviation styles may be defined using:

---

---

where ⟨name⟩ is the name of the new style (as used in the mandatory argument of \setabbreviationstyle). This is similar but not identical to the glossaries package’s \newacronymstyle command.

The ⟨setup⟩ argument deals with the way the entry is defined and may set attributes for the given abbreviation category. This argument should redefine

---

---

to set the entry fields including the name (defaults to the short form if omitted), sort, first, firstplural. Other fields may also be set, such as text, plural and description.

For example, the long-short style has the following in ⟨setup⟩:

  \renewcommand*{\CustomAbbreviationFields}{%
    name={\protect\glsabbrvfont{\the\glsshorttok}},
    sort={\the\glsshorttok},
    first={\protect\glsfirstlongfont{\the\glslongtok}%
     \protect\glsxtrfullsep{\the\glslabeltok}%
     \glsxtrparen{\protect\glsfirstabbrvfont{\the\glsshorttok}}},%
    firstplural={\protect\glsfirstlongfont{\the\glslongpltok}%
     \protect\glsxtrfullsep{\the\glslabeltok}%
     \glsxtrparen{\protect\glsfirstabbrvfont{\the\glsshortpltok}}},%
    plural={\protect\glsabbrvfont{\the\glsshortpltok}},%
    description={\the\glslongtok}}%

Note that the first and firstplural are set even though they’re not used by \gls.

The basic styles, such as long-short, use commands like \glsabbrvfont (which are redefined whenever the style formatting is set) within \CustomAbbreviationFields. Other styles, such as long-em-short-em directly use their own custom commands, such as \glsabbrvemfont. With these styles, commands like \glsabbrvfont still need to be defined as appropriate in the ⟨fmts⟩ argument even if they’re not used within \CustomAbbreviationFields.

The ⟨setup⟩ argument may also redefine

---

---

which can be used to assign attributes. (This will automatically be initialised to do nothing.)

For example, the short-footnote includes the following in ⟨setup⟩:

  \renewcommand*{\GlsXtrPostNewAbbreviation}{%
    \glssetattribute{\the\glslabeltok}{nohyperfirst}{true}%
    \glshasattribute{\the\glslabeltok}{regular}%
    {%
      \glssetattribute{\the\glslabeltok}{regular}{false}%
    }%
    {}%
  }%

This sets the nohyperfirst attribute to “true”. It also unsets the regular attribute if it has previously been set. Note that the nohyperfirst attribute doesn’t get unset by other styles, so take care not to switch styles for the same category.

You can access the short, long, short plural and long plural values through the following token registers.

Short value (defined by glossaries):

---

---

Short plural value (defined by glossaries-extra):

---

---

(This may be the default value or, if provided, the value provided by the user through the shortplural key in the optional argument of \newabbreviation.)

Long value (defined by glossaries):

---

---

Long plural value (defined by glossaries-extra):

---

---

(This may be the default value or, if provided, the value provided by the user through the longplural key in the optional argument of \newabbreviation.)

The short or long values may be modified by attributes (such as markwords). The above registers reflect the modification. If you want to access the original (unmodified) short or long form (as provided in the final two arguments of \newabbreviation), then use the commands:

---

---

for the short form and

---

---

for the long form. (These may be useful for the sort key to avoid any formatting that may be added by the attribute setting.)

There are two other registers available that are defined by glossaries:

---

---

which contains the entry’s label and

---

---

which contains the values provided in the optional argument of \newabbreviation.

Remember put \the in front of the register command as in the examples above. The category label can be access through the command (not a register):

---

---

This may be used inside the definition of \GlsXtrPostNewAbbreviation.

If you want to base a style on an existing style, you can use

---

---

where ⟨name⟩ is the name of the existing style. For example, the long-noshort-sc-desc style simply does

\GlsXtrUseAbbrStyleSetup{long-noshort-desc}

within ⟨setup⟩.

The ⟨fmts⟩ argument deals with the way the entry is displayed in the document. This argument should redefine the following commands.

The default suffix for the plural short form (if not overridden by the shortplural key):

---

---

(Note that this isn’t used for the plural long form, which just uses the regular \glspluralsuffix.)

The font used for the short form on first use or in the full forms:

---

---

The font used for the short form on subsequent use or through commands like \glsxtrshort:

---

---

The font used for the long form on first use or in the full forms:

---

---

The font used for the long form in commands like \glsxtrlong use:

---

---

Display full form singular no case-change (used by \gls on first use for abbreviations without the regular attribute set):

---

---

Display full form singular first letter converted to upper case (used by \Gls on first use for abbreviations without the regular attribute set):

---

---

Display full form plural no case-change (used by \glspl on first use for abbreviations without the regular attribute set):

---

---

Display full form plural first letter converted to upper case (used by \Glspl on first use for abbreviations without the regular attribute set):

---

---

In addition ⟨fmts⟩ may also redefine the following commands that govern the inline full formats. If the style doesn’t redefine them, they will default to the same as the display full forms.

Inline singular no case-change (used by \glsentryfull, \glsxtrfull and \GLSxtrfull):

---

---

Inline singular first letter converted to upper case (used by \Glsentryfull and \Glsxtrfull):

---

---

Inline plural no case-change (used by \glsentryfullpl, \glsxtrfullpl and \GLSxtrfullpl):

---

---

Inline plural first letter converted to upper case (used by \Glsentryfullpl and \Glsxtrfullpl):

---

---

(New to version 1.17.) You can also modify the way the subsequent use is formatted by redefining the following four commands, but these won’t be used for abbreviations with the regular attribute set. If the style doesn’t redefine these commands, the default values are used.

Singular with no case-change:

---

---

Singular with first letter upper case:

---

---

Plural with no case-change:

---

---

Plural with first letter upper case:

---

---

If you want to provide support for glossaries-accsupp use the following \glsaccess⟨xxx⟩ commands (§13.2 Accessibility Support) within the definitions of \glsxtrfullformat etc instead of the analogous \glsentry⟨xxx⟩ commands. (If you don’t use glossaries-accsupp, they will just do the corresponding \glsentry⟨xxx⟩ command.)

For example, the short-long style has the following in ⟨fmts⟩:

  \renewcommand*{\abbrvpluralsuffix}{\glsxtrabbrvpluralsuffix}%
  \renewcommand*{\glsabbrvfont}[1]{\glsabbrvdefaultfont{##1}}%
  \renewcommand*{\glsfirstabbrvfont}[1]{\glsfirstabbrvdefaultfont{##1}}%
  \renewcommand*{\glsfirstlongfont}[1]{\glsfirstlongdefaultfont{##1}}%
  \renewcommand*{\glslongfont}[1]{\glslongdefaultfont{##1}}%
  \renewcommand*{\glsxtrfullformat}[2]{%
    \glsfirstabbrvfont{\glsaccessshort{##1}\ifglsxtrinsertinside##2\fi}%
    \ifglsxtrinsertinside\else##2\fi
    \glsxtrfullsep{##1}%
    \glsxtrparen{\glsfirstlongfont{\glsaccesslong{##1}}}%
  }%
  \renewcommand*{\glsxtrfullplformat}[2]{%
    \glsfirstabbrvfont{\glsaccessshortpl{##1}\ifglsxtrinsertinside##2\fi}%
    \ifglsxtrinsertinside\else##2\fi
    \glsxtrfullsep{##1}%
    \glsxtrparen{\glsfirstlongfont{\glsaccesslongpl{##1}}}%
  }%
  \renewcommand*{\Glsxtrfullformat}[2]{%
    \glsfirstabbrvfont{\Glsaccessshort{##1}\ifglsxtrinsertinside##2\fi}%
    \ifglsxtrinsertinside\else##2\fi\glsxtrfullsep{##1}%
    \glsxtrparen{\glsfirstlongfont{\glsaccesslong{##1}}}%
  }%
  \renewcommand*{\Glsxtrfullplformat}[2]{%
    \glsfirstabbrvfont{\Glsaccessshortpl{##1}\ifglsxtrinsertinside##2\fi}%
     \ifglsxtrinsertinside\else##2\fi\glsxtrfullsep{##1}%
    \glsxtrparen{\glsfirstlongfont{\glsaccesslongpl{##1}}}%
  }%

Since the inline full commands aren’t redefined, they default to the same as the display versions.

If you want to base a style on an existing style, you can use

---

---

within ⟨fmts⟩, where ⟨name⟩ is the name of the existing style. For example, the long-short-desc style has the following in ⟨fmts⟩:

  \GlsXtrUseAbbrStyleFmts{long-short}%

Here’s an example of an abbreviation style that’s based on long-short that displays the short form within \textsf:

\newabbreviationstyle
{custom-sf}% label
{% setup
  \GlsXtrUseAbbrStyleSetup{short-long}%
}%
{% fmts
  \GlsXtrUseAbbrStyleFmts{short-long}%
  \renewcommand*{\glsabbrvfont}[1]{\textsf{##1}}%
}

Note that this wouldn’t work if it was instead based on one of the modified versions such as short-sc-long as they explicitly use their own formatting commands, such as \glsabbrvemfont. The base styles, such as short-long, use the more generic \glsabbrvfont etc which makes them easier to adapt than the modified styles.

For further details, see the “Abbreviations” section in the documented code (glossaries-extra-code.pdf).