latex-dev 2022-11 (2nov22)

git-svn-id: svn://tug.org/texlive/trunk@64899 c570f23f-e606-0410-a88d-b1316a301751

1 files changed, 164 insertions, 59 deletions

diff --git a/Master/texmf-dist/doc/latex-dev/base/ltnews36.tex b/Master/texmf-dist/doc/latex-dev/base/ltnews36.tex
index 55de98628cf..d6df801db2a 100644
--- a/Master/texmf-dist/doc/latex-dev/base/ltnews36.tex
+++ b/Master/texmf-dist/doc/latex-dev/base/ltnews36.tex
@@ -123,7 +123,7 @@
 \tubcommand{\input{tubltmac}}
 
 \publicationmonth{November}
-\publicationyear{2022  --- DRAFT version for upcoming release}
+\publicationyear{2022} %  --- DRAFT version for upcoming release}
 
 \publicationissue{36}
 
@@ -143,16 +143,30 @@
 
 \section{Introduction}
 
-\emph{to be written}
+The 2022-11 release of \LaTeX{} is largely a consolidation release
+where we made a number of minor improvements to fix some bugs or
+improve one or the other interface.
 
+The only really important functionality that was added is described in
+the next section: the ability to easily define document-level commands
+and environments that accept a key/value list in one of its (usually
+optional) arguments, including the ability to determine if the argument
+does in fact contain such a key/value list or just a single
+\enquote{classical} value.
 
-\subsection{Auto-detecting key--value arguments}
+For the \enquote{Tagged \LaTeX{} Project} this functionality is very
+important because many document-level commands will need to accept
+such key/value lists, for example, to specify alternative text or
+overwrite default tagging if that becomes necessary in a document.
+
+
+\section{Auto-detecting key/value arguments}
 
 To allow extension of the core \LaTeX{} syntax, \pkg{ltcmd} now supports
-a \texttt{={...}} modifier when grabbing arguments. This modifier instructs
+a \texttt{={...}}\ modifier when grabbing arguments. This modifier instructs
 \LaTeX{} that the argument should be passed to the underlying code as
-a set of keyvals. If the argument does not \enquote{look like} a set
-of keyvals, it will be converted into a single key--value pair, with
+a set of key/values. If the argument does not \enquote{look like} a set
+of key/values, it will be converted into a single key/value pair, with
 the argument to \texttt{=} specifying the name of that key. For
 example, the \cs{caption} command could be defined as
 \begin{verbatim}
@@ -161,26 +175,29 @@ example, the \cs{caption} command could be defined as
          {...}
 \end{verbatim}
 which would mean that if the optional argument does \emph{not}
-contain keyval data, it will be converted to a single keyval
+contain key/value data, it will be converted to a single key/value
 pair with the key name \texttt{short-text}.
 
 Arguments which begin with \texttt{=,} are always interpreted as
-keyvals even if they do not contain further \texttt{=} signs.
+key/values even if they do not contain further \texttt{=} signs.
 Any \texttt{=} signs enclosed within \verb|$...$| or \verb|\(...\)|,
 i.e.~in inline math mode, are ignored, meaning that
 only \texttt{=} outside of math mode will generally cause
-interpretation as keyval material.
+interpretation as key/value material.
 
 In case the argument contains a \enquote{textual} \texttt{=} sign that
-is mistaken as key/value indicator you can hide it using a brace
+is mistaken as a key/value indicator you can hide it using a brace
 group as you would do in other places, e.g.,
 \begin{verbatim}
 \caption[{Use of = signs}]
         {Use of = signs in optional arguments}
 \end{verbatim}
-However, because a \texttt{=} sign in math mode are already ignored, this
+However, because \texttt{=} signs in math mode are already ignored, this
 should seldom be necessary.
 
+
+\section{A note for font package developers}
+
 \subsection{Encoding subsets for \texttt{TS1} encoded fonts}
 
 The text companion encoding \texttt{TS1} is unfortunately not very
@@ -204,29 +221,51 @@ glyphs that are actually available in the font.\footnote{The \LaTeX{}
 %
 \githubissue{905}
 
-
-
 \section{New or improved commands}
 
+\subsection{Better language handling for case-changing commands}
+
+The commands \cs{MakeUppercase}, \cs{MakeLowercase} and \cs{MakeTitlecase} now
+automatically detect the locale currently in use when \pkg{babel} is loaded.
+This allows automatic adjustment of letter mappings where appropriate. They
+also accept a leading optional argument. This accepts a key--value list of
+control settings. At present, there is one key available: \texttt{locale},
+which can also be accessed via the alias \texttt{lang}. This is intended to
+allow local setting of the language, which can be done using a BCP-47
+descriptor. For example, this could be used to force Turkish case changing in
+otherwise English input
+\begin{verbatim}
+\MakeUppercase[lang = tr]{Ragıp Hulûsi Özdem}
+\end{verbatim}
+yields
+  \MakeUppercase[lang = tr]{Ragıp Hulûsi Özdem}.
+%   RAGIP HULÛS\.I ÖZDEM.                           %TUB
 
 \section{Code improvements}
 
-\subsection{Support for slanted small caps in the EC-fonts }
-Since some time \LaTeX{} supports the combination of the shapes
-small caps and italic/slanted. The EC-fonts contain slanted small caps fonts
+\subsection{Support for slanted small caps in the EC fonts}
+For some time \LaTeX{} has supported the combination of the shapes
+small caps and italic/slanted. The EC fonts contain slanted small caps fonts
 but using them required the loading of an external package. Suitable font definitions
 have now been added to \pkg{t1cmd.fd} and so from now on
 \begin{verbatim}
- \textsc{\textsl{slanted small}}
- \textsc{\textit{italic small caps}}
+\usepackage[T1]{fontenc}
+...
+ \textsc{\textsl{Slanted Small Caps}};
+ \textsc{\textit{Italic Small Caps}};
  \bfseries
- \textsc{\textsl{bold slanted small caps}}
- \textsc{\textit{bold italic small caps}}
+ \textsc{\textsl{Bold Slanted Small Caps}};
+ \textsc{\textit{Bold Italic Small Caps}}.
 \end{verbatim}
 will give the expected result: {\fontfamily{cmr}
-\textsc{\textsl{slanted small}} \textsc{\textit{italic small caps}}
+\textsc{\textsl{Slanted Small Caps}}; \textsc{\textit{Italic Small Caps}};
 \bfseries
-\textsc{\textsl{bold slanted small caps}} \textsc{\textit{bold italic small caps}}}
+\textsc{\textsl{Bold Slanted Small Caps}}; \textsc{\textit{Bold Italic Small Caps}}}.
+
+Given that the Computer Modern fonts in \texttt{T1} do not have real italic small caps but only
+slanted small caps, the latter is substituted for the former, which is
+why above both work but you see no difference between the two (and in the log you get a
+substitution warning for the \cs{textit}\cs{textsc} shape combination).
 %
 \githubissue{782}
 
@@ -242,28 +281,56 @@ font instead.
 \githubissue{879}
 
 
+
+\subsection{Improve font series handling with incorrect \texttt{.fd} files}
+
+By convention, the font series value is supposed to contain no
+\texttt{m}, unless you refer to the \enquote{medium} series (which is
+represented by a single \texttt{m}). For example, one should write
+\texttt{c} for \enquote{medium weight, condensed width} and not
+\texttt{mc}. This was one of the many space-conserving methods
+necessary in the early days of \LaTeXe.
+
+Some older \texttt{.fd} files
+do not obey that convention but use \texttt{mc}, \texttt{bm}, etc., in
+their declarations. As a result, some font selection scheme
+functionality was not working when confronted with such \texttt{.fd}
+files. We have therefore augmented \cs{DeclareSymbolFont} and
+\cs{SetSymbolFont} to strip any surplus \texttt{m} from their series argument
+so that they do not unnecessarily trigger font
+substitutions. Regardless of this support such 
+\texttt{.fd} files should get fixed by their maintainers.
+%
+\githubissue{918}
+
+
+
+
 \subsection{Detect nested \texttt{minipage} environments}
 
 Nesting of \texttt{minipage} environments is only partially supported
 in \LaTeX{} and can lead to incorrect output, such as overfull boxes
 or footnotes appearing in the wrong place;
 see~\cite[p.~106]{36:Lamport}. However, until now there was no warning
-if that happpened.  This has been changed and the environment now
+if that happened.  This has been changed and the environment now
 warns if you nest it in another \texttt{minipage} environment that
 already contains footnotes.
 %
 \githubissue{168}
 
 
-
-\subsection{\LuaTeX\ callback efficiency improvement}
-
-The mechanism for providing the
-\texttt{pre/post\_mlist\_to\_hlist\_filter} callbacks in \LuaTeX\ has
-been improved to make it more reusable and to avoid overhead if these
-callbacks are not used.
+\subsection{Robust commands in package options}
+With the standard key-based option handler added in the last release,
+or with contributed packages offering similar features, users may
+expect to be able to use a package option such as
+\verb|[font=\bfseries]|.
+Previously this failed with internal errors as the option list was
+expanded via \verb|\edef|. This has now been changed to use the
+existing command \verb|\protected@edef| so that any \LaTeX\ robust
+command should be safe to pass to a key value option.
 %
-\githubissue{830}
+\githubissue{932}
+
 
 
 
@@ -282,35 +349,56 @@ approach.
 
 
 
-\subsection{Improve font series handling with incorrect \texttt{.fd} files}
-
-By convention, the font series value is supposed to contain no
-\texttt{m}, unless you refer to the \enquote{medium} series (which is
-represented by a single \texttt{m}). For example, one should write
-\texttt{c} for \enquote{medium weight, condensed width} and not
-\texttt{mc}. This was one of the many space-conserving methods
-necessary in the early days of \LaTeXe.
+\subsection{\LuaTeX\ callback efficiency improvement}
 
-Some older \texttt{.fd} files
-do not obey that convention but use \texttt{mc}, \texttt{bm}, etc., in
-their declarations. As a result, some font selection scheme
-functionality was not working when confronted with such \texttt{.fd}
-files. We have therefore augmented \cs{DeclareSymbolFont} and
-\cs{SetSymbolFont} to strip their series argument from any surplus
-\texttt{m} so that they do not unnecessarily trigger font
-substitutions. Regardless of this support such 
-\texttt{.fd} files should get fixed by their maintainers.
+The mechanism for providing the
+\texttt{pre/post\_mlist\_to\_hlist\_filter} callbacks in \LuaTeX\ has
+been improved to make it more reusable and to avoid overhead if these
+callbacks are not used.
 %
-\githubissue{918}
+\githubissue{830}
+
+
+\subsection{Rule-based ordering for \LuaTeX\ callback handlers}
+
+In \hologo{LuaLaTeX} the callback handlers used to be called in the order
+in which they were registered in, but this was often rather fragile.
+It depends a lot on the load order and any attempts to enforce a
+different order required unregistering and reregistering the handlers to
+be reordered. Additionally, even if some ordering constraints where
+enforced that way, another package loaded later could accidentally
+overwrite it.
+
+To improve this, we now order the callback handlers based on ordering
+rules similar to the hook rules.
+
+When registering a callback which should run before or after another
+callback, \verb+luatexbase.declare_callback_rule+ can now be used to
+record this ordering constraint.
+For example
+\begin{verbatim}
+luatexbase.add_to_callback
+ ('pre_shaping_filter', my_handler, 'my_name')
+luatexbase.declare_callback_rule
+ ('pre_shaping_filter',
+            'my_name', 'before', 'other_name')
+\end{verbatim}
+will ensure that \verb+my_handler+ will always be called before the
+handler registered as \verb+other_name+.
+
+This also means that the order in which callbacks are registered no
+longer implicitly defines an order.
+Code which relied on this implicit order should now define the order
+rules explicitly.
 
 
 
 \section{Bug fixes}
 
-\subsection{Prevent \TeX{} from loosing a \cs{smash}}
+\subsection{Prevent \TeX{} from losing a \cs{smash}}
 
 When \TeX{} is typesetting a fraction, it will rebox the material in
-either numerator or the denumerator depending on which is wider. If
+either the numerator or denominator, depending on which is wider. If
 the repackaged part consists of a single box, that box gets new
 dimensions and if it was built using a \cs{smash} that effect vanishes
 (because a smash is nothing other than zeroing some box dimension,
@@ -321,28 +409,36 @@ which now got undone). For example, in the line
 \end{verbatim}
 the $2$ in the denominators was not always at the same vertical position, because
 the second \cs{smash} was ignored due to reboxing:
+\vspace{-1.2\baselineskip}
 \[
 \makeatletter
 \def\mathsm@sh#1#2{\setbox\z@\hbox{$\m@th#1{#2}$}\finsm@sh} % old definition
 \makeatother
 \newcommand*\drawbaseline{\rlap{\vrule width 60pt height 0.1pt depth 0pt }}
+\qquad
 \frac{1}{\drawbaseline2} = \frac{1}{\smash{2^X}}      \neq 
 \frac{100}{\smash{2^X}}
+\pagebreak     % not TUB
 \]
-The differences are subtle but noticeable. This is now corrected and the \cs{smash} is always
+The differences are subtle but noticeable.
+This is now corrected and the \cs{smash} is always
 honored. Thus now you get this output:
+\vspace{-1.2\baselineskip}
 \[
 \newcommand*\drawbaseline{\rlap{\vrule width 60pt height 0.1pt depth 0pt }}
+\qquad
 \frac{1}{\drawbaseline2} = \frac{1}{\smash{2^X}}      \neq 
 \frac{100}{\smash{2^X}}
 \]
+\par
+\vspace{-1.5\baselineskip}
 %
 \githubissue{517}
 
 
 \subsection{Resolve an issue with \cs{mathchoice} and \texttt{localalphabets}}
 
-The code for keeping a number of math alphabetcs local (introduced in
+The code for keeping a number of math alphabets local (introduced in
 2021; see~\cite{36:ltnews34}) used \cs{aftergroup} to do some cleanup actions after a
 formula had finished.  Unfortunately, \cs{aftergroup} can't be used
 inside the arguments of the \cs{mathchoice} primitive and as a result one
@@ -352,19 +448,28 @@ approach altogether.
 %
 \githubissue{921}
 
+\subsection{Reporting of unused global options when using key/value processing}
+
+Using the new key/value option processor did not properly report any unused
+global options when it was used in handling class options. This has now been
+corrected.
+%
+\githubissue{938}
 
 
-\section{Changes to packages in the \pkg{amsmath} category}
+%\section{Changes to packages in the \pkg{amsmath} category}
 
 
+%\newpage   TUB?
+
 \section{Changes to packages in the \pkg{graphics} category}
 
 \subsection{Fix a \cs{mathcolor} bug}
 
-The \cs{mathcolor} command intorduced in \cite{36:ltnews35} needs to
-scan for following sub and superscripts, but if it did so at the end
+The \cs{mathcolor} command introduced in \cite{36:ltnews35} needs to
+scan for following sub- and superscripts, but if it did so at the end
 of an alignment cell, e.g., in a \texttt{array} environment, the
-\texttt{\&} was evaluated too early causing some internal errors. This
+\texttt{\&} was evaluated too early, causing some internal errors. This
 is now properly guarded for.
 %
 \githubissue{901}
@@ -375,7 +480,7 @@ is now properly guarded for.
 
 \subsection{\pkg{array}: Correctly identify single-line m-cells}
 
-Cells in m-columns that only contain a single line are supposed to
+Cells in m-columns that contain only a single line are supposed to
 behave like single-line p-cells and align at the same baseline. To
 test for the condition, \pkg{array} used to compare the height of the cell to
 the height of the strut used for the table rows. However, the height of that
@@ -388,7 +493,7 @@ tall, in which case aligning is pointless anyway).
 \githubissue{766}
 
 
-\medskip
+%\medskip
 
 \begin{thebibliography}{9}