Debian Bug report logs - #24307
tetex-bin: jadetex causes TeX capacity exceeded

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: "Adam P. Harris" <apharris@burrito.onshore.com>

To: quade@lpr.e-technik.tu-muenchen.de

Cc: submit@bugs.debian.org

Subject: tetex-bin: jadetex causes TeX capacity exceeded

Date: Tue, 07 Jul 1998 10:38:09 -0400

Package: tetex-bin
Version: 0.9-7

Juergen Quade <quade@lpr.e-technik.tu-muenchen.de> writes:
> Find attached an archiv with the necessary files to reproduce the
> bug.  I stripped this down from my normal text. I have no problems
> generating html or rtf, but TeX is unfortunately not working.

> I produced this bug with several other "conditions" (base
> texts). But I think it is always the same bug, therefore it should
> be enough to provide you with this environment.

> Just unpack the archive and type make. I have a standard Debian
> system 2.0 and the newest (dist/slink/) jade, jadetex and tetex
> versions.

> If you have problems reproducing the bug or if you need some more
> questions, don't hesitate to ask me.

Juergen, I had no problem reproducing the bug:

  jadetex Jadetest.tex
  This is TeX, Version 3.14159 (Web2C 7.2)
  (Jadetest.tex
  LaTeX2e <1997/12/01> patch level 2
  Babel <v3.6j> and hyphenation patterns for american, french, german, nohyphenat
  ion, loaded.
  (/usr/lib/texmf/tex/latex/psnfss/t1ptm.fd)
  JADE TeX starts
  Elements will be labelled
  [...]
  [3]
  ! TeX capacity exceeded, sorry [save size=4000].
  \SetupICs ...csname Posture@\fPosture \endcsname }
                                \edef \f@family {\expandaf...
  l.530 {}
        \Seq%

It's very consistent.

> Thanks very much for tracking down the problem,

Well, here's hoping!

Christoph, did you get the files to reproduce the bug?  Is this a
jadetex bug rather than a tetex-bin bug?

.....A. P. Harris...apharris@onShore.com...<URL:http://www.onShore.com/>

-- System Information
Debian Release: 2.0
Kernel Version: Linux burrito 2.0.34 #1 Sun Jun 14 14:43:30 EDT 1998 i586 unknown

Versions of the packages tetex-bin depends on:
ii  libc6           2.0.7r-2       The GNU C library version 2 (run-time files)
ii  libpng0g        0.96-5         PNG library - runtime
ii  xlib6g          3.3.2.2-1      shared libraries required by X clients
ii  zlib1g          1.1.1-0.1      compression library - runtime
ii  tetex-base      0.9-9          basic teTeX library files
ii  ed              0.2-16         The classic unix line editor
ii  dpkg-perl       0.1-2hamm1     Perl interface modules for dpkg

--- Begin /etc/texmf/texmf.cnf (modified conffile)
% original texmf.cnf -- runtime path configuration file for kpathsea.
% (If you change or delete `original' on the previous line, the
% distribution won't install its version over yours.)
% Public domain.
% 
% What follows is a super-summary of what this .cnf file can
% contain. Please read the Kpathsea manual for more information.
% 
% texmf.cnf is generated from texmf.in, by replacing @var@ with the
% value of the Make variable `var', via a sed file texmf.sed, generated
% (once) by kpathsea/Makefile (itself generated from kpathsea/Makefile.in
% by configure).
% 
% Any identifier (sticking to A-Za-z_ for names is safest) can be assigned.
% The `=' (and surrounding spaces) is optional.
% No % or @ in texmf.in, for the sake of autogeneration.
% (However, %'s and @'s can be edited into texmf.cnf or put in envvar values.)
% $foo (or ${foo}) in a value expands to the envvar or cnf value of foo.
% 
% Earlier entries (in the same or another file) override later ones, and
% an environment variable foo overrides any texmf.cnf definition of foo.
% 
% All definitions are read before anything is expanded, so you can use
% variables before they are defined. 
% 
% If a variable assignment is qualified with `.PROGRAM', it is ignored
% unless the current executable (last filename component of argv[0]) is
% named PROGRAM.  This foo.PROGRAM construct is not recognized on the
% right-hand side. For environment variables, use FOO_PROGRAM.
% 
% Which file formats use which paths for searches is described in the
% various programs' and the kpathsea documentation.
% 
% // means to search subdirectories (recursively).
% A leading !! means to look only in the ls-R db, never on the disk.
% A leading/trailing/doubled : in the paths will be expanded into the
%   compile-time default. Probably not what you want.
% 
% You can use brace notation, for example: /usr/local/{mytex,othertex}
% expands to /usr/local/mytex:/usr/local/othertex.  Instead of the path
% separator you can use a comma: /usr/local/{mytex,othertex} also expands
% to /usr/local/mytex:/usr/local/othertex.  However, the use of the comma
% instead of the path separator is deprecated.
%
% The text above assumes thet path separator is a colon (:).  Non-UNIX
% systems use different path separators, for example MicroSoft operating
% systems use the semicolon (;).
%  Part 1: Search paths and directories.
% You can set an environment variable to override TEXMF if you're testing
% a new TeX tree, without changing anything else.
% 
% You may wish to use one of the $SELFAUTO... variables here so TeX will
% find where to look dynamically.  See the manual and the definition
% below of TEXMFCNF.
% The main tree, which must be mentioned in $TEXMF, below:
TEXMFMAIN = /usr/lib/texmf
% A place for local additions to a "standard" texmf tree.  For example:
TEXMFLOCAL = /usr/lib/texmf/local
% User texmf trees can be catered for like this...
HOMETEXMF = $HOME/texmf
% Now, list all the texmf trees. If you have multiple trees,
% use shell brace notation, like this:
TEXMF = {$HOMETEXMF,!!$TEXMFLOCAL,!!$TEXMFMAIN}
% The braces are necessary.
%TEXMF = !!$TEXMFMAIN
% The system trees.  These are the trees that are shared by all the users.
SYSTEXMF = $TEXMF
% Where generated fonts may be written.  This tree is used when the sources
% were found in a system tree and either that tree wasn't writable, or the
% varfonts feature was enabled in MT_FEATURES in mktex.cnf.
VARTEXFONTS  = /var/spool/texmf
% Where to look for ls-R files.  There need not be an ls-R in the
% directories in this path, but if there is one, Kpathsea will use it.
TEXMFDBS = $TEXMF:$VARTEXFONTS
% It may be convenient to define TEXMF like this:
%   TEXMF = {$HOMETEXMF,!!$TEXMFLOCAL,!!$TEXMFMAIN,$HOME}
% which allows users to set up entire texmf trees, and tells TeX to
% look in places like ~/tex and ~/bibtex.  If you do this, define TEXMFDBS
% like this:
%   TEXMFDBS = $HOMETEXMF;$TEXMFLOCAL;$TEXMFMAIN;$VARTEXFONTS
% or mktexlsr will generate an ls-R file for $HOME when called, which is
% rarely desirable.  If you do this you'll want to define SYSTEXMF like
% this:
%   SYSTEXMF = $TEXMFLOCAL;$TEXMFMAIN
% so that fonts from a user's tree won't escape into the global trees.
%
% On some systems, there will be a system tree which contains all the font
% files that may be created as well as the formats.  For example
%   VARTEXMF = /var/lib/texmf
% is used on many Linux systems.  In this case, set VARTEXFONTS like this
%   VARTEXFONTS = $VARTEXMF/fonts
% and do not mention it in TEXMFDBS (but _do_ mention VARTEXMF).
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Usually you will not need to edit any of the other variables in part 1. %
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% WEB2C is for Web2C specific files.  The current directory may not be
% a good place to look for them.
WEB2C = $TEXMF/web2c
% TEXINPUTS is for TeX input files -- i.e., anything to be found by \input
% or \openin, including .sty, .eps, etc. 
% LaTeX 2e specific macros are stored in latex/, macros that can only be
% used with 2.09 in latex209/.  In addition, we look in the directory
% latex209, useful for macros that were written for 2.09 and do not
% mention 2e at all, but can be used with 2e.
TEXINPUTS.latex209 = .:$TEXMF/tex/{latex209,generic,latex,}//
TEXINPUTS.latex2e = .:$TEXMF/tex/{latex,generic,latex209,}//
TEXINPUTS.latex = .:$TEXMF/tex/{latex,generic,latex209,}//
% Fontinst needs to read afm files.
TEXINPUTS.fontinst = .:$TEXMF/{tex,fonts/afm}//
% Plain TeX.  Have the command tex check all directories as a last
% resort, we may have plain-compatible stuff anywhere.
TEXINPUTS.tex = .:$TEXMF/tex/{plain,generic,}//
% other plain-based formats
TEXINPUTS.amstex = .:$TEXMF/tex/{amstex,plain,generic,}//
TEXINPUTS.ftex = .:$TEXMF/tex/{formate,plain,generic,}//
TEXINPUTS.texinfo = .:$TEXMF/tex/{texinfo,plain,generic,}//
TEXINPUTS.eplain  = .:$TEXMF/tex/{eplain,plain,generic,}//
TEXINPUTS.jadetex = .:$TEXMF/tex/{jadetex,plain,generic,}//
% MLTeX.
TEXINPUTS.frtex   = .:$TEXMF/{mltex,tex}/{french,plain,generic,}//
TEXINPUTS.frlatex = .:$TEXMF/{mltex,tex}/{french,latex,generic,latex209,}//
TEXINPUTS.mltex = .:$TEXMF/{mltex,tex}/{plain,generic,}//
TEXINPUTS.mllatex = .:$TEXMF/{mltex,tex}/{latex,generic,latex209,}//
% e-TeX.  This form of the input paths is borrowed from teTeX.  A certain
% variant of TDS is assumed here, unaffected by the build variables.
TEXINPUTS.elatex   = .:$TEXMF/{etex,tex}/{latex,generic,latex209,plain,}//
TEXINPUTS.etex     = .:$TEXMF/{etex,tex}/{plain,generic,}//
% PDFTeX.  This form of the input paths is borrowed from teTeX.  A certain
% variant of TDS is assumed here, unaffected by the build variables.
TEXINPUTS.pdftexinfo = .:$TEXMF/{pdftex,tex}/{texinfo,generic,plain,}//
TEXINPUTS.pdflatex = .:$TEXMF/{pdftex,tex}/{latex,generic,latex209,plain,}//
TEXINPUTS.pdftex   = .:$TEXMF/{pdftex,tex}/{plain,generic,}//
% Omega.
TEXINPUTS.lambda = .:$TEXMF/{omega,tex}/{lambda,latex,generic,latex209,plain,}//
TEXINPUTS.omega = .:$TEXMF/{omega,tex}/{plain,generic,}//
% Earlier entries override later ones, so put this last.
TEXINPUTS = .:$TEXMF/tex/{generic,}//
% Metafont, MetaPost inputs.
MFINPUTS = .:$TEXMF/metafont//:{$TEXMF/fonts,$VARTEXFONTS}/source//
MPINPUTS = .:$TEXMF/metapost//
% Dump files (fmt/base/mem) for vir{tex,mf,mp} to read (see web2c/INSTALL),
% and string pools (.pool) for ini{tex,mf,mp}.  It is silly that we have six
% paths and directories here (they all resolve to a single place by default),
% but historically ...
TEXFORMATS = .:$TEXMF/web2c
MFBASES = .:$TEXMF/web2c
MPMEMS = .:$TEXMF/web2c
TEXPOOL = .:$TEXMF/web2c
MFPOOL = .:$TEXMF/web2c
MPPOOL = .:$TEXMF/web2c
% Device-independent font metric files.
VFFONTS = .:$TEXMF/fonts/vf//
TFMFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/tfm//
% The $MAKETEX_MODE below means the drivers will not use a cx font when
% the mode is ricoh.  If no mode is explicitly specified, kpse_prog_init
% sets MAKETEX_MODE to /, so all subdirectories are searched.  See the manual.
% The modeless part guarantees that bitmaps for PostScript fonts are found.
PKFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/pk/{$MAKETEX_MODE,modeless}//
% Similarly for the GF format, which only remains in existence because
% Metafont outputs it (and MF isn't going to change).
GFFONTS = .:$TEXMF/fonts/gf/$MAKETEX_MODE//
% A backup for PKFONTS and GFFONTS. Not used for anything.
GLYPHFONTS = .:$TEXMF/fonts
% For texfonts.map and included map files used by mktexpk.
% See ftp://ftp.tug.org/tex/fontname.tar.gz.
TEXFONTMAPS = .:$TEXMF/fontname
% BibTeX bibliographies and style files.
BIBINPUTS = .:$TEXMF/bibtex/{bib,}//
BSTINPUTS = .:$TEXMF/bibtex/{bst,}//
% MFT style files.
MFTINPUTS = .:$TEXMF/mft//
% PostScript headers, prologues (.pro), encodings (.enc) and fonts.
TEXPSHEADERS.pdftex     = .:$TEXMF/{pdftex,dvips,fonts/type1}//
TEXPSHEADERS.pdflatex   = .:$TEXMF/{pdftex,dvips,fonts/type1}//
TEXPSHEADERS.pdftexinfo = .:$TEXMF/{pdftex,dvips,fonts/type1}//
TEXPSHEADERS = .:$TEXMF/{dvips,pdftex,fonts/type1}//
% PostScript Type 1 outline fonts.
T1FONTS = .:$TEXMF/fonts/type1//
% PostScript AFM metric files.
AFMFONTS = .:$TEXMF/fonts/afm//
% TrueType outline fonts.
TTFONTS = .:$TEXMF/fonts/truetype//
% Type 42 outline fonts.
T42FONTS = .:$TEXMF/fonts/type42//
% Dvips' config.* files (this name should not start with `TEX'!).
TEXCONFIG = .:$TEXMF/dvips//
% Makeindex style (.ist) files.
INDEXSTYLE = .:$TEXMF/makeindex//
% Used by DMP (ditroff-to-mpx), called by makempx -troff.
TRFONTS = /usr/lib/font/devpost
MPSUPPORT = .:$TEXMF/metapost/support
% For xdvi to find mime.types and .mailcap, if they do not exist in
% $HOME.  These are single directories, not paths.
% (But the default mime.types, at least, may well suffice.)
MIMELIBDIR = /usr/etc
MAILCAPLIBDIR = /usr/etc
% TeX documentation and source files, for use with kpsewhich.
TEXDOCS = .:$TEXMF/doc//
TEXSOURCES = .:$TEXMF/source//
% Omega-related fonts and other files.  The odd construction for OFMFONTS
% makes it behave in the face of a definition of TFMFONTS.  Unfortunately
% no default substitution would take place for TFMFONTS, so an explicit
% path is retained.
OFMFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/{ofm,tfm}//:$TFMFONTS
OPLFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/opl//
OVFFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/ovf//
OVPFONTS = .:{$TEXMF/fonts,$VARTEXFONTS}/ovp//
OTPINPUTS = .:$TEXMF/omega/otp//
OCPINPUTS = .:$TEXMF/omega/ocp//
% Some additional input variables for several programs.  If you add
% a program that uses the 'other text files' or 'other binary files'
% search formats, you'll want to add their variables here as well.
XDVIINPUTS = .:$TEXMF/xdvi
TEX4HTINPUTS = .:$TEXMF/tex4ht//
%% The mktex* scripts rely on KPSE_DOT. Do not set it in the environment.
KPSE_DOT = .
% This definition isn't used from this .cnf file itself (that would be
% paradoxical), but the compile-time default in paths.h is built from it.
% The SELFAUTO* variables are set automatically from the location of
% argv[0], in kpse_set_progname.  
% 
% About the /. construction:
% 1) if the variable is undefined, we'd otherwise have an empty path
%    element in the compile-time path.  This is not meaningful.
% 2) if we used /$VARIABLE, we'd end up with // if VARIABLE is defined,
%    which would search the entire world.
%    
% The TETEXDIR stuff isn't likely to relevant unless you're using teTeX,
% but it doesn't hurt.
% 
TEXMFCNF = .:{$SELFAUTOLOC,$SELFAUTODIR,$SELFAUTOPARENT}{,{/share,}/texmf{.local,}/web2c}:/.$TETEXDIR:/.$TEXMF/web2c:/usr/lib/texmf/web2c
%  Part 2: Non-path options.
 
% Write .log/.dvi/etc. files here, if the current directory is unwritable.
% TEXMFOUTPUT = /tmp
% If a dynamic file creation fails, log the command to this file, in
% either the current directory or TEXMFOUTPUT.  Set to the
% empty string or  0  to avoid logging.
MISSFONT_LOG = missfont.log
% Set to a colon-separated list of words specifying warnings to suppress.
% To suppress everything, use TEX_HUSH = all; this is equivalent to
% TEX_HUSH = checksum:lostchar:readable:special
TEX_HUSH = none
% Enable system commands via \write18{...}?
shell_escape = f
% Allow TeX \openout on filenames starting with `.' (e.g., .rhosts)?
% a (any)        : any file can be opened.
% r (restricted) : disallow opening "dotfiles".
% p (paranoid)   : as 'r' and disallow going to parent directories, and
%                  restrict absolute paths to be under $TEXMFOUTPUT.
openout_any = p
% Enable the mktex... scripts by default?  These must be set to 0 or 1.
% Particular programs can and do override these settings, for example
% dvips's -M option.  Your first chance to specify whether the scripts
% are invoked by default is at configure time.
% 
% These values are ignored if the script names are changed; e.g., if you
% set DVIPSMAKEPK to `foo', what counts is the value of the environment
% variable/config value `FOO', not the `MKTEXPK' value.
% 
% MKTEXTEX = 0
% MKTEXPK = 0
% MKTEXMF = 0
% MKTEXTFM = 0
% MKOCP = 0
% MKOFM = 0
% What MetaPost runs to make MPX files.  This is passed an option -troff
% if MP is in troff mode.  Set to `0' to disable this feature.
MPXCOMMAND = makempx
%  Part 3: Array and other sizes for TeX (and Metafont and MetaPost).
% 
% If you want to change some of these sizes only for a certain TeX
% variant, the usual dot notation works, e.g.,
% main_memory.hugetex = 20000000
% 
% If a change here appears to be ignored, try redumping the format file.
% Memory. Must be less than 8,000,000 total.
% 
% main_memory is relevant only to initex, extra_mem_* only to non-ini.
% Thus, have to redump the .fmt file after changing main_memory; to add
% to existing fmt files, increase extra_mem_*.  (To get an idea of how
% much, try \tracingstats=2 in your TeX source file;
% web2c/tests/memtest.tex might also be interesting.)
% 
% To increase space for boxes (as might be needed by, e.g., PiCTeX),
% increase extra_mem_bot.
% 
% For some xy-pic samples, you may need as much as 700000 words of memory.
% For the vast majority of documents, 60000 or less will do.
% 
main_memory = 263000 % words of inimemory available: also applies to inimf&mp
extra_mem_top = 0    % extra high memory for chars, tokens, etc.
extra_mem_bot = 0    % extra low memory for boxes, glue, breakpoints, etc.
% Words of font info for TeX (total size of all TFM files, approximately). 
font_mem_size = 200000
% Total number of fonts. Must be >= 50 and <= 2000 (without tex.ch changes).
font_max = 1000
% Extra space for the hash table of control sequences (which allows 10K
% names as distributed).
hash_extra = 0
% Max number of characters in all strings, including all error messages,
% help texts, font names, control sequences.  These values apply to TeX and MP.
pool_size = 125000		
% Minimum pool space after TeX/MP's own strings; must be at least
% 25000 less than pool_size, but doesn't need to be nearly that large.
string_vacancies = 25000
max_strings = 15000		% max number of strings
pool_free = 5000		% min pool space left after loading .fmt
% Hyphenation trie. As distributed, the maximum is 65535; this should
% work unless `unsigned short' is not supported or is smaller than 16
% bits.  This value should suffice for UK English, US English, French,
% and German (for example).  To increase, you must change
% `ssup_trie_opcode' and `ssup_trie_size' in tex.ch (and rebuild TeX);
% the trie will then consume four bytes per entry, instead of two.
% 
% US English, German, and Portuguese: 30000.
% German: 14000.
% US English: 10000.
% 
trie_size = 64000
% Buffer size.  TeX uses the buffer to contain input lines, but macro
% expansion works by writing material into the buffer and reparsing the
% line.  As a consequence, certain constructs require the buffer to be
% very large.  As distributed, the size is 50000; most documents can be
% handled within a tenth of this size.
buf_size = 50000
hyph_size = 1000        % number of hyphenation exceptions, >610 and <32767.
nest_size = 100		% simultaneous semantic levels (e.g., groups)
max_in_open = 15	% simultaneous input files and error insertions
param_size = 500	% simultaneous macro parameters
save_size = 4000	% for saving values outside current group
stack_size = 300	% simultaneous input sources
% These are Omega-specific.
ocp_buf_size = 20000	% character buffers for ocp filters.
ocp_stack_size = 10000	% stacks for ocp computations.
ocp_list_size = 1000	% control for multiple ocps.
% These work best if they are the same as the I/O buffer size, but it
% doesn't matter much.  Must be a multiple of 8.
dvi_buf_size = 16384 % TeX
gf_buf_size = 16384  % MF
% It's probably inadvisable to change these. At any rate, we must have:
% 45 < error_line      < 255;
% 30 < half_error_line < error_line - 15;
% 60 <= max_print_line;
% These apply to Metafont and MetaPost as well.
error_line = 79
half_error_line = 50
max_print_line = 79

--- End /etc/texmf/texmf.cnf

--- Begin /etc/texmf/mktex.cnf (modified conffile)
: ${MT_FEATURES=appendonlydir:varfonts}

--- End /etc/texmf/mktex.cnf

Message #10 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Christoph Martin <martin@uni-mainz.de>

To: "Adam P. Harris" <apharris@burrito.onshore.com>, 24307@bugs.debian.org

Cc: quade@lpr.e-technik.tu-muenchen.de, Christoph.Martin@uni-mainz.de

Subject: Re: Bug#24307: tetex-bin: jadetex causes TeX capacity exceeded

Date: Tue, 7 Jul 1998 17:37:37 +0200 (MESZ)

Adam P. Harris writes:
 >   ! TeX capacity exceeded, sorry [save size=4000].
 >   \SetupICs ...csname Posture@\fPosture \endcsname }
 >                                 \edef \f@family {\expandaf...
 >   l.530 {}
 >         \Seq%
 > 
 > It's very consistent.
 > 
 > > Thanks very much for tracking down the problem,
 > 
 > Well, here's hoping!
 > 
 > Christoph, did you get the files to reproduce the bug?  Is this a

Yes, I got the files.

 > jadetex bug rather than a tetex-bin bug?

I will look into this asap.

Christoph

Message #15 received at 24307@bugs.debian.org (full text, mbox, reply):

From: apharris@burrito.onshore.com (Adam P. Harris)

To: 24307@bugs.debian.org

Subject: followup to tetex capacity exceeed

Date: 29 Jul 1998 19:57:20 -0400

Christoph, any update on this bug yet?
-- 
.....A. P. Harris...apharris@onShore.com...<URL:http://www.onShore.com/>

Message #20 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Christoph Martin <martin@uni-mainz.de>

To: apharris@burrito.onshore.com (Adam P. Harris), 24307@bugs.debian.org

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Thu, 30 Jul 1998 10:24:19 +0200 (MET DST)

Adam P. Harris writes:
 > 
 > Christoph, any update on this bug yet?

Sorry, I had no time yet to look into it :/

Christoph

Message #25 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Antti-Juhani Kaijanaho <antkaij@rolle.cc.jyu.fi>

To: Christoph Martin <martin@uni-mainz.de>, 24307@bugs.debian.org, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Thu, 30 Jul 1998 12:19:43 +0300

On Thu, Jul 30, 1998 at 10:24:19AM +0200, Christoph Martin wrote:
> Adam P. Harris writes:
>  > 
>  > Christoph, any update on this bug yet?
> 
> Sorry, I had no time yet to look into it :/

Sorry for the intrusion...

I just read what the BTS has on this bug.  The problem seems to be
that one of TeX's static arrays has become full.  It is rare that this
means anything else than that the original file has an error.  However,
you can easily check this:

There is a config file, texmf.cnf (probably in /etc on Debian systems,
I'm not on one right now so I can't check) that controls, among other
things, the size of the static arrays in TeX.  Look for the following:

8<--- excerpt
% Buffer size.  TeX uses the buffer to contain input lines, but macro
% expansion works by writing material into the buffer and reparsing the
% line.  As a consequence, certain constructs require the buffer to be
% very large.  As distributed, the size is 50000; most documents can be
% handled within a tenth of this size.
buf_size = 50000

hyph_size = 1000        % number of hyphenation exceptions, >610 and <32767.
nest_size.context = 500
nest_size = 100         % simultaneous semantic levels (e.g., groups)
max_in_open = 15        % simultaneous input files and error insertions
param_size.context = 1500
param_size = 500        % simultaneous macro parameters
save_size.context = 5000
save_size = 4000        % for saving values outside current group
stack_size.context = 1500
stack_size = 300        % simultaneous input sources
8<---

Try this: enlarge save_size and retry processing the problemous file.
You may need to redump the formats before trying to process (TeX will
complain if this is necessary) - ie do 'texconfig init' as root; you
may have to do something special to redump the jadetex format.
Iterate until you have no problems.  If this does not help, the bug is
in the input file.  Otherwise, the TeX Debian ships is too small for
processing the file.

IMHO, in any case it is a misfeature of Jade, if not a bug, that it
consumes so much of TeX's memory.



        Antti-Juhani
-- 
Antti-Juhani Kaijanaho <gaia@iki.fi> ** <URL:http://www.iki.fi/gaia/> **

         I can't seem to find a lowercase 'r' on my keyboard.
        (Lee Davies in comp.unix.programmer on July 22, 1998)

Message #30 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

To: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Thu, 30 Jul 1998 12:37:39 +0200

On Thu, Jul 30, 1998 at 12:19:43PM +0300, Antti-Juhani Kaijanaho wrote:
> On Thu, Jul 30, 1998 at 10:24:19AM +0200, Christoph Martin wrote:
> > Adam P. Harris writes:
> >  > 
> >  > Christoph, any update on this bug yet?
> > 
> > Sorry, I had no time yet to look into it :/
> 
> IMHO, in any case it is a misfeature of Jade, if not a bug, that it
> consumes so much of TeX's memory.

I had this when using hundreds of ID Attributes in the docbook document (a
tutorial I write with autogenerated reference pages).

I had to increase three buffers, but then it worked fine.

Marcus

-- 
"Rhubarb is no Egyptian god."        Debian GNU/Linux        finger brinkmd@ 
Marcus Brinkmann                   http://www.debian.org    master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de                        for public  PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       PGP Key ID 36E7CD09

Message #35 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Antti-Juhani Kaijanaho <antkaij@rolle.cc.jyu.fi>

To: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Fri, 31 Jul 1998 01:46:49 +0300

On Thu, Jul 30, 1998 at 12:37:39PM +0200, Marcus Brinkmann wrote:
> I had this when using hundreds of ID Attributes in the docbook document (a
> tutorial I write with autogenerated reference pages).

It seems that you have a TeX-wise unusually complex document.  I don't
think the distributed version needs enlargement, but a note in the
docs about the feature (new in teTeX 0.9, BTW) could do good.

BTW, the author of Web2C - on which teTeX is based - has said that he
might implement true dynamic memory in a future version.  If he does
that, the next teTeX version (probably some 1.x) might need no
tweaking of conffiles with unusually complex inputs.



        Antti-Juhani
-- 
Antti-Juhani Kaijanaho <gaia@iki.fi> ** <URL:http://www.iki.fi/gaia/> **

         I can't seem to find a lowercase 'r' on my keyboard.
        (Lee Davies in comp.unix.programmer on July 22, 1998)

Message #40 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

To: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Fri, 31 Jul 1998 16:54:13 +0200

On Fri, Jul 31, 1998 at 01:46:49AM +0300, Antti-Juhani Kaijanaho wrote:
> On Thu, Jul 30, 1998 at 12:37:39PM +0200, Marcus Brinkmann wrote:
> > I had this when using hundreds of ID Attributes in the docbook document (a
> > tutorial I write with autogenerated reference pages).
> 
> It seems that you have a TeX-wise unusually complex document.  I don't
> think the distributed version needs enlargement, but a note in the
> docs about the feature (new in teTeX 0.9, BTW) could do good.

Well, the document is automatically generated from SGML source. The
resulting document is only 60 pages. Maybe Jade TeX output does some strange
things, should be worth to investigate. How does LaTeX scale with a great
amount of global identifiers (for references)...

> BTW, the author of Web2C - on which teTeX is based - has said that he
> might implement true dynamic memory in a future version.  If he does
> that, the next teTeX version (probably some 1.x) might need no
> tweaking of conffiles with unusually complex inputs.

Sounds cool. However, tweaking the conffiles work, which means that there is
probably not a leak in the JadeTeX files, but only an inefficient way to allocate
memory.

Thank you,
Marcus

-- 
"Rhubarb is no Egyptian god."        Debian GNU/Linux        finger brinkmd@ 
Marcus Brinkmann                   http://www.debian.org    master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de                        for public  PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       PGP Key ID 36E7CD09

Message #45 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Antti-Juhani Kaijanaho <antkaij@rolle.cc.jyu.fi>

To: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Fri, 31 Jul 1998 18:26:46 +0300

On Fri, Jul 31, 1998 at 04:54:13PM +0200, Marcus Brinkmann wrote:
> On Fri, Jul 31, 1998 at 01:46:49AM +0300, Antti-Juhani Kaijanaho wrote:
> > I don't
> > think the distributed version needs enlargement, but a note in the
> > docs about the feature (new in teTeX 0.9, BTW) could do good.

I need to rephrase this, as the above is confusing: "I don't think it
is necessary to enlarge the buffers in the shipped Debian package of
teTeX.  However, there should be a note in the docs about the
possibility of enlarging the buffers by modifying the config files."

> Well, the document is automatically generated from SGML source. The
> resulting document is only 60 pages. 

Strange.  A modern TeX should have no problems processing a 60-page
document.

> Maybe Jade TeX output does some strange things,

This is likely.

Can you send me one of the TeX source files (as produced by Jade) by
private email (I don't think the BTS would like one)?

Some backround for all this: TeX does all its memory management by
preallocating some static arrays and then doing its own housekeeping
within them.  Don Knuth wrote TeX vesrion 2 (aka TeX82) in 1982, and
he believed back then that Pascal's new is unportable.  The TeX 3
revision in 1989 was only a revision for making TeX 8-bit clean, and
memory management was not touched.  Nowadays TeX is frozen, so Knuth's
pristine sources will never have true dynamic memory.

Currently there are only two TeX distributions that rewrite TeX's
memory management in order to use true dynamic memory, but they both
are commercial, and neither of them is available for Unices.

> How does LaTeX scale with a great amount of global identifiers (for
> references)...

Well, people write their theses with LaTeX with no problems.  Recently
someone complained about a legit "TeX capacity exceeded" message in
comp.text.tex, but the book being produced was several thousand pages
long...

In general however, if you see a "TeX capacity exceeded", look for
errors in the input.

> Sounds cool. However, tweaking the conffiles work, which means that
> there is probably not a leak in the JadeTeX files, but only an
> inefficient way to allocate memory.

It is possible that Jade is wasting TeX's memory by using
unnecessarily long reference keys.



        Antti-Juhani
-- 
Antti-Juhani Kaijanaho <gaia@iki.fi> ** <URL:http://www.iki.fi/gaia/> **

         I can't seem to find a lowercase 'r' on my keyboard.
        (Lee Davies in comp.unix.programmer on July 22, 1998)

Message #50 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

To: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Fri, 31 Jul 1998 17:36:25 +0200

On Fri, Jul 31, 1998 at 06:26:46PM +0300, Antti-Juhani Kaijanaho wrote:
> On Fri, Jul 31, 1998 at 04:54:13PM +0200, Marcus Brinkmann wrote:
> > On Fri, Jul 31, 1998 at 01:46:49AM +0300, Antti-Juhani Kaijanaho wrote:
> 
> Can you send me one of the TeX source files (as produced by Jade) by
> private email (I don't think the BTS would like one)?

I can do this, but I will do some investigation on my own, to isolate it a
bit (because jadetex uses an own macro package, and I'm alos interested in
this).
 
> > Sounds cool. However, tweaking the conffiles work, which means that
> > there is probably not a leak in the JadeTeX files, but only an
> > inefficient way to allocate memory.
> 
> It is possible that Jade is wasting TeX's memory by using
> unnecessarily long reference keys.

Mmmh. I used long id names (they were generated from class name, member name
and function argument types), the id values were up to 80 or more characters
:) Maybe this is a good answer for this. If it is so, I think Jade should
map the sgml id's to shorter unique latex id's.

Thank you for your explanations,
Marcus

-- 
"Rhubarb is no Egyptian god."        Debian GNU/Linux        finger brinkmd@ 
Marcus Brinkmann                   http://www.debian.org    master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de                        for public  PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       PGP Key ID 36E7CD09

Message #55 received at 24307@bugs.debian.org (full text, mbox, reply):

From: "Adam P. Harris" <apharris@burrito.onshore.com>

To: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

Cc: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org, Christoph Martin <martin@uni-mainz.de>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Fri, 31 Jul 1998 12:14:22 -0400

Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> writes:
> On Fri, Jul 31, 1998 at 06:26:46PM +0300, Antti-Juhani Kaijanaho
> wrote:
>> On Fri, Jul 31, 1998 at 04:54:13PM +0200, Marcus Brinkmann wrote:
>> > On Fri, Jul 31, 1998 at 01:46:49AM +0300, Antti-Juhani Kaijanaho
>> > wrote:
>>
>> Can you send me one of the TeX source files (as produced by Jade)
>> by private email (I don't think the BTS would like one)?

> I can do this, but I will do some investigation on my own, to
> isolate it a bit (because jadetex uses an own macro package, and I'm
> alos interested in this).
 
I have the original SGML which can always make it crash (haven't
gotten a chance to try tweaking tex variables in /etc yet).  Would
that help in tracking down what jadetex is doing?

And yes, I share the suspicion that jadetex is doing something very
counter-productive...

.....A. P. Harris...apharris@onShore.com...<URL:http://www.onShore.com/>

Message #60 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Juergen Quade <Juergen.Quade@lpr.e-technik.tu-muenchen.de>

To: 24307@bugs.debian.org

Cc: quade@spock.lpr.e-technik.tu-muenchen.de (Juergen Quade), apharris@onShore.com

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Sat, 1 Aug 1998 13:38:42 +0200 (CEST)

Increasing the memory buffers of TeX (/etc/texmf/texmf.cnf) helped.
I am now able to process all my Jadetex generated files.

Thanks a lot for the workaround!

         Jürgen.

Message #65 received at 24307@bugs.debian.org (full text, mbox, reply):

From: apharris@burrito.onshore.com (Adam P. Harris)

To: control@bugs.debian.org

Cc: 24307@bugs.debian.org

Subject: tetex-bin: jadetex causes TeX capacity exceeded

Date: 03 Sep 1998 13:21:34 -0400

reassign 24307 jadetex
thanks

I am refiling this bug against jadetex.  I think it really is a
jadetex "bug", and all we really need to do to fix it is mention the
workaround (contained in this bug log) in
/usr/doc/jadetex/README.Debian or something.

.....A. P. Harris...apharris@onShore.com...<URL:http://www.onShore.com/>

Message #72 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Antti-Juhani Kaijanaho <antkaij@oskari.cc.jyu.fi>

To: 24307@bugs.debian.org

Cc: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>, Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Bug#24307: followup to tetex capacity exceeed

Date: Mon, 14 Sep 1998 18:37:49 +0300

Hi,

David Carlisle recently posted in comp.text.tex another workaround
which similar to my original workaround for this problem.  Carlisle's
original message can be viewed at <URL:http://www.dejanews.com/
getdoc.xp?AN=390109397>.  His solution should probably be implemented
in teTeX, not in Jade.



        Antti-Juhani
-- 
Antti-Juhani Kaijanaho <gaia@iki.fi> ** <URL:http://www.iki.fi/gaia/> **

          All GNU users have more.  Most of them have less.
                       Some of them have most.

Message #77 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

To: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org

Cc: Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Mon, 14 Sep 1998 19:42:50 +0200

On Mon, Sep 14, 1998 at 06:37:49PM +0300, Antti-Juhani Kaijanaho wrote:
> Hi,
> 
> David Carlisle recently posted in comp.text.tex another workaround
> which similar to my original workaround for this problem.  Carlisle's
> original message can be viewed at <URL:http://www.dejanews.com/
> getdoc.xp?AN=390109397>.  His solution should probably be implemented
> in teTeX, not in Jade.

But on the other hand, I think that JadeTeX could save some ressources on
its own. I suspect that one problem are the labels for references. If
JadeTeX would enumerate the labels and just use a small number instead of
the whole original label, some ressources could be spared.

Thank oyu,
Marcus

-- 
"Rhubarb is no Egyptian god."        Debian GNU/Linux        finger brinkmd@ 
Marcus Brinkmann                   http://www.debian.org    master.debian.org
Marcus.Brinkmann@ruhr-uni-bochum.de                        for public  PGP Key
http://homepage.ruhr-uni-bochum.de/Marcus.Brinkmann/       PGP Key ID 36E7CD09

Message #82 received at 24307@bugs.debian.org (full text, mbox, reply):

From: Antti-Juhani Kaijanaho <antkaij@oskari.cc.jyu.fi>

To: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>, 24307@bugs.debian.org

Cc: Christoph Martin <martin@uni-mainz.de>, "Adam P. Harris" <apharris@burrito.onshore.com>

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Mon, 14 Sep 1998 20:53:01 +0300

On Mon, Sep 14, 1998 at 07:42:50PM +0200, Marcus Brinkmann wrote:
> But on the other hand, I think that JadeTeX could save some ressources on
> its own. 

Agreed.  In the meantime, however, Carlisle's solution is clean enough
to be implemented in the teTeX package, i.e. it does not alter the
limits for other formats, just for jadetex.  Somebody should find
working limits first, however.


        Antti-Juhani
-- 
Antti-Juhani Kaijanaho <gaia@iki.fi> ** <URL:http://www.iki.fi/gaia/> **

          All GNU users have more.  Most of them have less.
                       Some of them have most.

Message #87 received at 24307@bugs.debian.org (full text, mbox, reply):

From: "Adam P. Harris" <apharris@burrito.onshore.com>

To: Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de>

Cc: Antti-Juhani Kaijanaho <gaia@iki.fi>, 24307@bugs.debian.org

Subject: Re: Bug#24307: followup to tetex capacity exceeed

Date: Mon, 14 Sep 1998 14:10:04 -0400

Marcus Brinkmann <Marcus.Brinkmann@ruhr-uni-bochum.de> writes:
> On Mon, Sep 14, 1998 at 06:37:49PM +0300, Antti-Juhani Kaijanaho
> wrote:
>> Hi,
>> 
>> David Carlisle recently posted in comp.text.tex another workaround
>> which similar to my original workaround for this problem.
>> Carlisle's original message can be viewed at
>> <URL:http://www.dejanews.com/getdoc.xp?AN=390109397>.  His
>> solution should probably be implemented in teTeX, not in Jade.

> But on the other hand, I think that JadeTeX could save some
> ressources on its own. I suspect that one problem are the labels for
> references. If JadeTeX would enumerate the labels and just use a
> small number instead of the whole original label, some ressources
> could be spared.

I quite agree that at heart it is a jadetex problem.  Jadetex puts
excessive requirements on TeX due to the rather perverse way it uses
TeX.  I'm quite sure that the Jadetex upstream author is aware of this
defect and has it on the burner (probably near the back, it might be
quite hard to fix).  I believe this bug should be put into forwarded
state, with a mention of the problem/workaround made in the
documentation to jadetex.

.....A. P. Harris...apharris@onShore.com...<URL:http://www.onShore.com/>

Send a report that this bug log contains spam.