path: root/support/glosstex/TODO

$Id: TODO,v 1.66 1997/12/03 18:08:22 volkan Exp $

* BUGS

** the .gdf-files to be used should be designated within the document
   and so put into the .aux-file, not designated on the commandline.
   The same way as it is done with BibTeX and \bibliography{foo,
   bar}. Problem: this would break the UI.

* `FEATURES'

** A Makeindex input line cannot exceed 1024 characters on one side,
   but an argument cannot span more than one line. Seems as though we
   should have a look at xindy. Another way around this is to change
   the makeindex sources (#define ARGUMENT_MAX in mkind.h) and
   recompile. This should be noted in the documentation of glosstex.

** while scanning the input (.aux)-files, glosstex is limited to an
   arbitrary number of characters per entry (constant argument to
   sscanf). this can be circumvented by using the GNU version of
   sscanf, which automatically malloc's the needed memory.

** LaTeX produces a nasty error `! LaTeX Error: Something's
   wrong--perhaps a missing \item.' when the .glx-file is empty.

* TODO

** \GLX@output@short and \GLX@output@long both should get called with
   <paren> and <form> to allow for greater flexibility. Or
   alternatively, provide \GLX@output@short@lt, lf, it, if. That way,
   we wouldn't need to parse the <form>-stuff. Just pass <paren>.

** document all the `.' and `o' that are output.

** An "Die TeXnische Komödie" article about GlossTeX would be nice.

** add support for xindy

* MAYBE

** include functionality of nomencl, à la include references produced
   by \ref and not only \pageref, or eq-ref. problem: too much
   features. it's not easy to deterimine the environment we're
   currently in. Also, nomencl and glosstex complement each other
   well, don't they?

** include a test-suite with files foo.glx.expect to be able to test
   each version during development whether it's still fully
   functional. that'd be hard to do, since we expect DVI-files, don't
   we? we may produce some expected GXS or GLX files.

** each list should be able to produce independent lists on
   part/chapter/section-level. problem: hard to implement, featurism,
   not really necessary.

** it should be possible to make a list with the terms sorted in order
   of appearance. We wouldn't need MakeIndex in this case. This may
   lead to a future release which doesn't rely on MakeIndex
   anymore. Something like BibTeX which has it's own language for it's
   .bst-files. Or reimplement GlossTeX using a .bst for BibTeX.

* DONE

** make sure that the following lines don't appear in glosstex.mst
   (which btw should really be called glosstex.ist)

   \def\filename{glosstex}
   \def\fileversion{0.4a}
   \def\filedate{1997/06/12}
   \let\docdate=\filedate

** TDS: glosstex.mst should be called glosstex.ist (which is more
   common in the TeX world) and moved to texmf/makeindex

** Damit die Acronym-expansion funktionieren kann, muß die .glx-Datei
   mindestens einmal abgearbeitet werden, auch wenn gar keine Liste
   erwünscht ist. Abhilfe: \printglosstex(foo) oder neuen Befehl
   \printglosstex* einführen, der keine Ausgabe erzeugt.

   Put something into \AtEndOfPackage{\printglosstex{foobarkazong}} or
   better, with \AtBeginDocument

** use README.m4 as source for README and LICENSE

** ist ein begriff explizit in irgendeiner liste explizit angefordert,
   wird dieser aber per "*" etwa in einer anderen liste auch
   angefordert, taucht er dort nicht auf. möglichkeit: wildcards
   werden in einem 2. pass eingetragen, da explizite referenzen
   priorität haben. 1.37 ist halbwegs korrekte version, bei der dieser
   bug noch auftritt.

** the endmark \glosstexpage := \hfill\page doesn't work reliably. see
   thmmarks.sty (I can't seem to find it) and ntheorem.sty for a
   better solution. Or avoid this problem at all. There are also some
   minor layout problems due to formatting the output.

** was genau passiert, wenn irgendwelche einträge mit "*" angefordert
   werden, dieselben einträge aber auch explizit? in welchen listen
   tauchen die ganzen einträge dann auf? Das sollte genauestens
   Dokumentiert werden, auch im Source!

   Die Sache ist erledigt. Alles läuft jetzt in 2 passes! Änderungen
   v.a. in database.c


** (from sad@utk.edu) Die Fu"snotenidee mag ich. Die Package option
   auch, aber vielleicht w"urde ich es dennoch als zwei getrennt
   erreichbare Versionen implementieren, also z.B.

 \acf        (entweder als ' (<item>) <long-form>' oder als
             '<long-form> (<item>)', je nach package option, und
	     entweder im text oder als footnote, je nach package option)
 \acft       (entweder als ' (<item>) <long-form>' oder 
             '<long-form> (<item>)', je nach package option, aber im
	     text)
 \acftl      (immer als '<long-form> (<item>)' aber im text)
 \acfti      (immer als '(<item>) <long-form>' aber im text)
 \acff       (wie \acft{} aber als footnote)
 \acffl      (wie oben, aber als footnote)
 \acffi      (wie oben, aber als footnote)


Das gibt die gr"o"ste Freiheit f"ur's Formatieren, denn manchmal klingt,
vor allem im Deutschen mit diesen W"ortern, die zwei Kapitel spannen und
immer noch dreizeilige Acronyme produzieren, '<long-form> (<item>)'
einfach traurig, vor allem, wenn 'long-form' irgendwelche grammatischen
Schn"orkel  beinhaltet. Hm, Beispiel, an den Haaren herbeigezogen:

 Wir waren in  'Die nordamerikanische Tomatenschlacht' (DNATS)

klingt trauriger als das meiner Meinung nach bessere jedoch 
sich dem glosstex Ansatz entziehende

 Wir waren in  der 'nordamerikanischen Tomatenschlacht' (DNATS)

w"ahrend 

 Wir waren in  DNATS ('Die nordamerikanische Tomatenschlacht')

das umgeht ohne den Filmtitel zu ver"andern.

Da f"allt dann noch eine andere M"oglichkeit f"ur optionen a la
Makeindex auf: Wahlweise 'open quote' 'close quote' Symbole um die
Langform, sagen wir mal per default als blanks wenn vorangestellt und
als '(' und ')' wenn hinten an, aber einstellbar als alles m"ogliche,
z.~B. als  die folgenden Paare (einzeln ansprechbar)

  ' '
  " "
  ` `
  ` '
  `` ''
  ( )
  [ ]
  \{ \}
  -- --
  --- ---
  , ,

oder auch beliebige andere folgen, wie 

 (' ')
 --- oder ausf"uhrlicher geschreiben,    , selbst wenn sich das ziemlich
 beschissen liest ---


usw. :-) Dann vielleicht noch ein bi"schen Feinkontrolle "uber's
Erscheinungsbild des Textes (sagen wir mal, \textbf{}. \textsc{}.
\text{tt} usw) getrennt f"ur beide Teile, und alle interessanten
M"oglichkeiten sollten ersch"opft sein. Ich f"ande das ganz sch"on, auch
wenn man das nat"urlich schon itzo per Hand mit ein paar Macros machen
kann. (Nat"urlich sollte glosstex dann tcp/ip enabled sein und einen
report per sendmail zu Dir schicken, falls irgendwer au"ser mir
diese Schn"orkel wirklich benutzt :->).

** in \GLX@do@acf die Klammern () durch Macros ersetzen, abhängig von
   <list>

** removed little bug in database.c that caused failure while parsing
   a <long-form> containing a ``,``.

** there's a strange bug in the Makefile that causes on recursive
   invocation of $(MAKE) Makefile.os2 and Makefile.unx to be read
   additionally to Makefile. dev only. This should be worth a
   bug-report, but first construct this bug on a minimal makefile.

   Oh man, I am so stupid. MAKEFILES is a special variable in
   make. RTFM!

   Have a look at the Makefile from doc++ in /usr/local/lib/doc++/src,
   especially the "release" target.

** write a kpathsea sh-wrapper around glosstex.bin to be used with
   teTeX

   or

   call MakeIndex from within GlossTeX with exec(). Should also work
   with DOS (process.h instead of unistd.h in UNIX). name of MakeIndex
   binary should be configureable. in DOS, it's called makeindx,
   AFAIK.

   or

   some path searching functionality to find .gdf-files and .mst files
   (in combination with the above). maybe a sh-kpathsea-wrapper like
   makeindex and makeindex.bin in teTeX? Drawback: only functional
   under Unix (so what!). Another solution: combine this with the
   kpathsea sh-wrapper, something similar to texi2dvi, which also
   calls makeindex and so forth.

   note: kpse is some kind of black magic to me (at least at the
   moment), so the instructions below may be complete nonsense,
   nonstandard and absolutely ridiculous. but they work for me.

      check wether "kpsetool -v -n glosstex '$GDFINPUTS'" or "kpsetool
      -v -n tex '$GDFINPUTS'" do what you want, i.e. their output
      should be equal. These won't find your teTeX texmf tree (where
      your texmf.cnf file lives) if glosstex is not in something like
      teTeX/bin or "teTeX/bin/foo-bar". in these cases, replace
      "TETEXDIR = $SELFAUTOPARENT" in texmf.cnf with something like
      "TETEXDIR = /usr/local/lib/teTeX". this makes sure that the
      proper texmf tree is found, even if glosstex is not in the teTeX
      texmf tree but in your local TeX tree.

      You also need symbolic links from
      /usr/local/lib/localTeX/texmf.cnf to your teTeX texmf.cnf. This
      is because kpatsea looks for texmf.cnf in the tree where the
      binary was found.

** when glosstex is called with non-existant files, the output files
   .gxg and gxs are created nevertheless. Maybe we should unlink them
   before exiting?

** do some stuff with autoconf and automake to automatically generate
   glosstex-i486-unknown-linux binaries for distrib.

   done this with uname -something.

** parenl und parenr auch auf item-ebene festlegen. done using
   \ac,lparen,rparen,<form>(list)[pageref-mode]{item}

** \newcommand{\glosstexgroup} for use with lethead_prefix in
   MakeIndex, similar to \nomgroup. not really necessary. what is this
   good for, anyhow?

   This wont work, since glosstex produces list_item, to embed all
   items from all lists into one list for makeindex. makeindex would
   only recognize a new group based on the first letter or the list,
   which is not what we want.

** some macros have to be declared \long (\newcommand*) to be able to
   accept whole paragraphs as arguments. Especially those who define
   the begin environment stuff

** update documentation. new user-level commands:

   options: itemfirst, longfirst, text, footnote, roundparen,
   squareparen

   <i> : item/short-form first oder <l>: long-form first
   <t> : inline in text oder <g>: as footnote

   new commands: 
       \ac<form>(list)[pageref-mode]{item}
       \acf<form>(list)[pageref-mode]{item}
       
       glxitemorderdefault[list]{i|l}, glxitemplacementdefault[list]{t|f},
       glxparendefault[list]{left}{right}

** clean up the documentation:

   use more logical markup in documentation instead of visual. also
   make the documentation and source more consistent with respect to
   terminology ** check that at EOL, there is always a % in .dtx

   document each macro in dtx, especially the arguments

** move most of the info from the READMES into the main
   documentation. The README should only contain the most basic idea
   on what glosstex is.

** \AtEndOfDocument tell user, that there were undefined terms and
   that she should run glosstex.

** contact the maintainer of the latex catalogue

;;; Local Variables:
;;; mode: indented-text
;;; mode: outline-minor
;;; End: