Greetings ConTeXters, Just recently (after updating to the most recent i-Installer ConTeXt version / but also after writing/tweaking a little more), I am apparently running into some font memory issues.* I have tried to increase the available memory in web2c/texmf.cnf to no avail. One curiosity is the pdfTeX error further down: "<argument> hz,normal". I am using hz,hanging in this document. Any clues? Many thanks, David * [27.27] ! Font \*12ptrmtf*=texnansi-LIGA-KERN-ONUM-PNUM-MinionPro-Regular at 12.0pt not loaded: Not enough room left. <to be read again> \relax \xxdododefinefont ...tspec {#4}\newfontidentifier \let \localrelativefontsiz... \fontstrategy ...ame #1\csname #2#3#4#5\endcsname \tryingfontfalse \fi <inserted text> ...yle \fontalternative \fontsize \fi \iftryingfont \fontstr... \synchronizefont ...strategy \the \fontstrategies \relax \fi \ifskipfontchar... <argument> \getvalue {\@style@ \globalfontstyle } \edef \fontstyle {\globalfo... ... l.27 \stopbackmatter ! pdfTeX error (font expansion): invalid font identifier. \dosetfontadjusting ...pdffontexpand \handledfont \csname \@fha@ \@fha@ #1\c... <inserted text> ...ntadjusting \askedfonthandling \setprotrudingfactor \aske... \dofastenablehandling ...\everyenablefonthandling \edef \fonthandling {\csna... \rawprocesscommaitem ...commalevel \endcsname {#1} \expandafter \rawprocessco... <argument> hz,n ormal \rawprocesscommalist ...er \rawprocesscommaitem #1 ,],\global \advance \comma... ... l.27 \stopbackmatter ! ==> Fatal error occurred, the output PDF file is not finished!
Excuse me for replying to myself, but it is also curious that ConTeXt is trying to call a 12pt font. I have called 11pt in the environment. It looks like these errors are appearing at the Bibliography. Might this be due to something in the the beta bib-module? David On Jan 10, 2006, at 11:07 AM, David Wooten wrote:
Greetings ConTeXters,
Just recently (after updating to the most recent i-Installer ConTeXt version / but also after writing/tweaking a little more), I am apparently running into some font memory issues.*
I have tried to increase the available memory in web2c/texmf.cnf to no avail. One curiosity is the pdfTeX error further down: "<argument> hz,normal". I am using hz,hanging in this document.
Any clues?
Many thanks, David
* [27.27] ! Font \*12ptrmtf*=texnansi-LIGA-KERN-ONUM-PNUM-MinionPro-Regular at 12.0pt not loaded: Not enough room left. <to be read again> \relax \xxdododefinefont ...tspec {#4}\newfontidentifier \let \localrelativefontsiz...
\fontstrategy ...ame #1\csname #2#3#4#5\endcsname \tryingfontfalse \fi <inserted text> ...yle \fontalternative \fontsize \fi \iftryingfont \fontstr...
\synchronizefont ...strategy \the \fontstrategies \relax \fi \ifskipfontchar... <argument> \getvalue {\@style@ \globalfontstyle } \edef \fontstyle {\globalfo... ... l.27 \stopbackmatter
! pdfTeX error (font expansion): invalid font identifier. \dosetfontadjusting ...pdffontexpand \handledfont \csname \@fha@ \@fha@ #1\c... <inserted text> ...ntadjusting \askedfonthandling
\setprotrudingfactor \aske...
\dofastenablehandling ...\everyenablefonthandling \edef \fonthandling {\csna...
\rawprocesscommaitem ...commalevel \endcsname {#1} \expandafter \rawprocessco... <argument> hz,n ormal \rawprocesscommalist ...er \rawprocesscommaitem #1 ,],\global \advance \comma... ... l.27 \stopbackmatter
! ==> Fatal error occurred, the output PDF file is not finished! _______________________________________________ ntg-context mailing list ntg-context@ntg.nl http://www.ntg.nl/mailman/listinfo/ntg-context
David Wooten wrote:
Excuse me for replying to myself, but it is also curious that ConTeXt is trying to call a 12pt font. I have called 11pt in the environment.
It looks like these errors are appearing at the Bibliography. Might this be due to something in the the beta bib-module?
Not knowingly, but it could be the last drop. Can remove the \placepublications for a test run, and post the tail end of the log of that (hopefully succesful) run? The bit that contains the statistics, I mean. Cheers, taco
Here are the stats, after removing placepublications:
Here is how much of TeX's memory you used:
4760 strings out of 60999
76752 string characters out of 611011
32604822 words of memory out of 34111765
40674 multiletter control sequences out of 10000+50000
1974579 words of font info for 355 fonts, out of 2000000 for 2000
232 hyphenation exceptions out of 5000
62i,21n,81p,1198b,1736s stack positions out of 5000i,500n,6000p,200000b,40000s
PDF statistics:
209 PDF objects out of 300000
0 named destinations out of 131072
115969 words of extra memory for PDF output out of 128383
On Tue, 10 Jan 2006 21:22:14 +0100, Taco Hoekwater
David Wooten wrote:
Excuse me for replying to myself, but it is also curious that ConTeXt is trying to call a 12pt font. I have called 11pt in the environment.
It looks like these errors are appearing at the Bibliography. Might this be due to something in the the beta bib-module?
Not knowingly, but it could be the last drop.
Can remove the placepublications for a test run, and post the tail end of the log of that (hopefully succesful) run? The bit that contains the statistics, I mean.
Cheers, taco _______________________________________________ ntg-context mailing list ntg-context@ntg.nl http://www.ntg.nl/mailman/listinfo/ntg-context
David Wooten wrote:
Here are the stats, after removing placepublications:
Here is how much of TeX's memory you used: 4760 strings out of 60999 76752 string characters out of 611011 32604822 words of memory out of 34111765 40674 multiletter control sequences out of 10000+50000 1974579 words of font info for 355 fonts, out of 2000000 for 2000
That's cutting it close: 2000000-1974579 = 25421. Another five extra fonts (or so) being loaded will 'do it'. You have to increase font_mem_size, my guess is this limit is triggered by nothing but 'normal operation' (not a bug). Good luck, Taco
Taco Hoekwater wrote:
David Wooten wrote:
Here are the stats, after removing placepublications:
Here is how much of TeX's memory you used: 4760 strings out of 60999 76752 string characters out of 611011 32604822 words of memory out of 34111765 40674 multiletter control sequences out of 10000+50000 1974579 words of font info for 355 fonts, out of 2000000 for 2000
That's cutting it close:
2000000-1974579 = 25421.
Another five extra fonts (or so) being loaded will 'do it'.
You have to increase font_mem_size, my guess is this limit is triggered by nothing but 'normal operation' (not a bug).
indeed, it may be that the bibliography module uses a (say) small or slanted font and since context allocates fonts on demand (only when needed) such a problem may surface rather late in the file Hans
David Wooten wrote:
Greetings ConTeXters,
Just recently (after updating to the most recent i-Installer ConTeXt version / but also after writing/tweaking a little more), I am apparently running into some font memory issues.*
I have tried to increase the available memory in web2c/texmf.cnf to no avail. One curiosity is the pdfTeX error further down: "<argument> hz,normal". I am using hz,hanging in this document.
did you remake the format after increasing the memory values? hz lets pdftex internally generate additional font instances these are my texmf.cnf values buf_size.context = 200000 % needed for omega bug extra_mem_bot.context = 2000000 extra_mem_top.context = 4000000 font_max.context = 2000 font_mem_size.context = 1000000 hash_extra.context = 75000 main_memory.context = 2000000 max_strings.context = 250000 nest_size.context = 500 obj_tab_size.context = 300000 % 8388607 pdf_mem_size.context = 500000 % 524288 dest_names_size.context = 300000 % 131072 param_size.context = 10000 pool_free.context = 47500 pool_size.context = 2000000 save_size.context = 50000 stack_size.context = 10000 string_vacancies.context = 125000 trie_size.context = 250000 hyph_size.context = 8191 % prime (texmfstart --bin youreditor kpse:texmf.cnf) Hans
On Jan 10, 2006, at 1:54 PM, Hans Hagen wrote:
did you remake the format after increasing the memory values?
This means texexec --make --all ? I did try the above, but the numbers in the log when running my file do not change. I can only imagine that there is another place to change the memory settings. In my context.cnf, the numbers are identical to yours below, although I did try to raise the font.mem.size.context as we've been discussing.
hz lets pdftex internally generate additional font instances
these are my texmf.cnf values
buf_size.context = 200000 % needed for omega bug extra_mem_bot.context = 2000000 extra_mem_top.context = 4000000 font_max.context = 2000 font_mem_size.context = 1000000 hash_extra.context = 75000 main_memory.context = 2000000 max_strings.context = 250000 nest_size.context = 500 obj_tab_size.context = 300000 % 8388607 pdf_mem_size.context = 500000 % 524288 dest_names_size.context = 300000 % 131072 param_size.context = 10000 pool_free.context = 47500 pool_size.context = 2000000 save_size.context = 50000 stack_size.context = 10000 string_vacancies.context = 125000 trie_size.context = 250000 hyph_size.context = 8191 % prime
(texmfstart --bin youreditor kpse:texmf.cnf)
Hans
David Wooten wrote:
On Jan 10, 2006, at 1:54 PM, Hans Hagen wrote:
did you remake the format after increasing the memory values?
This means texexec --make --all ?
I did try the above, but the numbers in the log when running my file do not change. I can only imagine that there is another place to change the memory settings. In my context.cnf, the numbers are identical to yours below, although I did try to raise the font.mem.size.context as we've been discussing.
ah, you need to set the values in texmf.cnf! context.cnf is just an example (unless you point the TEXMFCNF environment variable to the alternative cnf file) Hans
On Jan 11, 2006, at 12:16 AM, Hans Hagen wrote:
David Wooten wrote:
On Jan 10, 2006, at 1:54 PM, Hans Hagen wrote:
did you remake the format after increasing the memory values?
This means texexec --make --all ?
I did try the above, but the numbers in the log when running my file do not change. I can only imagine that there is another place to change the memory settings. In my context.cnf, the numbers are identical to yours below, although I did try to raise the font.mem.size.context as we've been discussing.
ah, you need to set the values in texmf.cnf! context.cnf is just an example (unless you point the TEXMFCNF environment variable to the alternative cnf file)
Hans
Ah, unfortunately still no luck. I changed the settings in /usr/local/ teTeX/share/texmf/web2c/texmf.cnf, reset the format files with texconfig init, and for good measure texexec --make --all, but no change. Am I simply off target somewhere? Thanks, David
Find out which is the main texmf.cnf file: kpsewhich texmf.cnf on my system, this resolves to /usr/local/teTeX/texmf.cnf Then modify this file and regenerate the formats. HTH Thomas On Jan 12, 2006, at 12:17 AM, David Wooten wrote:
Ah, unfortunately still no luck. I changed the settings in /usr/ local/teTeX/share/texmf/web2c/texmf.cnf, reset the format files with texconfig init, and for good measure texexec --make --all, but no change. Am I simply off target somewhere?
Thanks, David
David Wooten wrote:
On Jan 11, 2006, at 12:16 AM, Hans Hagen wrote:
David Wooten wrote:
On Jan 10, 2006, at 1:54 PM, Hans Hagen wrote:
did you remake the format after increasing the memory values?
This means texexec --make --all ?
I did try the above, but the numbers in the log when running my file do not change. I can only imagine that there is another place to change the memory settings. In my context.cnf, the numbers are identical to yours below, although I did try to raise the font.mem.size.context as we've been discussing.
ah, you need to set the values in texmf.cnf! context.cnf is just an example (unless you point the TEXMFCNF environment variable to the alternative cnf file)
Hans
Ah, unfortunately still no luck. I changed the settings in /usr/local/ teTeX/share/texmf/web2c/texmf.cnf, reset the format files with texconfig init, and for good measure texexec --make --all, but no change. Am I simply off target somewhere?
maybe you have duplicate formats (texconfig does not support engine paths) do the reported mem values change? Hans
On Jan 11, 2006, at 11:25 PM, Thomas A. Schmitz wrote:
Find out which is the main texmf.cnf file:
kpsewhich texmf.cnf
on my system, this resolves to
/usr/local/teTeX/texmf.cnf
Then modify this file and regenerate the formats.
HTH
Thomas
Indeed, that is the case also for me.
maybe you have duplicate formats (texconfig does not support engine paths)
do the reported mem values change?
Hans
The reported values do not change, and afaik the formats are all listed here:
locate .fmt /usr/local/teTeX/share/texmf.local/web2c/amstex.fmt /usr/local/teTeX/share/texmf.local/web2c/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/cont-nl.fmt /usr/local/teTeX/share/texmf.local/web2c/eplain.fmt /usr/local/teTeX/share/texmf.local/web2c/etex.fmt /usr/local/teTeX/share/texmf.local/web2c/jadetex.fmt /usr/local/teTeX/share/texmf.local/web2c/latex.fmt /usr/local/teTeX/share/texmf.local/web2c/mptopdf.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-nl.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/mptopdf.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfjadetex.fmt /usr/local/teTeX/share/texmf.local/web2c/pdflatex.fmt /usr/local/teTeX/share/texmf.local/web2c/pdftex.fmt /usr/local/teTeX/share/texmf.local/web2c/pdftexinfo.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfxmltex.fmt /usr/local/teTeX/share/texmf.local/web2c/tex.fmt /usr/local/teTeX/share/texmf.local/web2c/texinfo.fmt /usr/local/teTeX/share/texmf.local/web2c/xmltex.fmt
Hi David, Old formats that should be deleted:
/usr/local/teTeX/share/texmf.local/web2c/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/cont-nl.fmt .. /usr/local/teTeX/share/texmf.local/web2c/mptopdf.fmt
New ones that have never been used yet:
/usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-nl.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/mptopdf.fmt
Cheers, Taco
On Jan 12, 2006, at 12:02 PM, Taco Hoekwater wrote:
Hi David,
Old formats that should be deleted:
/usr/local/teTeX/share/texmf.local/web2c/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/cont-nl.fmt .. /usr/local/teTeX/share/texmf.local/web2c/mptopdf.fmt
New ones that have never been used yet:
/usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-en.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/cont-nl.fmt /usr/local/teTeX/share/texmf.local/web2c/pdfetex/mptopdf.fmt
Cheers, Taco
Thanks Taco, Should those .../pdfetex/ fmt files remain where they are, and the .../web2c/ files simply deleted? David
David Wooten wrote:
On Jan 11, 2006, at 11:25 PM, Thomas A. Schmitz wrote:
Find out which is the main texmf.cnf file:
kpsewhich texmf.cnf
on my system, this resolves to
/usr/local/teTeX/texmf.cnf
Then modify this file and regenerate the formats.
HTH
Thomas
Indeed, that is the case also for me.
maybe you have duplicate formats (texconfig does not support engine paths)
do the reported mem values change?
Hans
The reported values do not change, and afaik the formats are all listed here:
can it be that you have multiple cnf files? kpse reads more than one if present Hans
On Jan 12, 2006, at 12:05 PM, Hans Hagen wrote:
can it be that you have multiple cnf files? kpse reads more than one if present
Here are all the .cnf files: /usr/local/teTeX/share/texmf/web2c/fmtutil.cnf /usr/local/teTeX/share/texmf/web2c/texmf.cnf /usr/local/teTeX/share/texmf.gwtex/texmf.cnf.gwtex.tl2003 /usr/local/teTeX/share/texmf.gwtex/texmf.cnf.gwtex.tl2004 /usr/local/teTeX/share/texmf.gwtex/web2c/mktex.cnf /usr/local/teTeX/share/texmf.local/web2c/context.cnf /usr/local/teTeX/share/texmf.local/web2c/fmtutil.cnf /usr/local/teTeX/share/texmf.tetex/web2c/mktex.cnf /usr/local/teTeX/texmf.cnf Even with multiple cnf files, if the memory settings are identical, should it work? Also, should the new memory settings have the ".context" suffix, e.g. font-mem-size.context? I believe I have tried both ways. Many thanks, David
Most Patient ConTeXters, I've rested a couple of days and found that, in fact, Gerben's distribution generates the new fmt files in ~/Library/texmf/web2c. Still, none of my attempts have produced changes in the allocated memory. Has anyone successfully increased the memory (e.g. font_mem_size) in this distribution? And if yes, could you say how it was achieved? Kind regards, David On Jan 12, 2006, at 2:48 PM, David Wooten wrote:
On Jan 12, 2006, at 12:05 PM, Hans Hagen wrote:
can it be that you have multiple cnf files? kpse reads more than one if present
Even with multiple cnf files, if the memory settings are identical, should it work?
Many thanks, David
No, that's not correct, it's not Gerben's distribution that does that. Which commands do you use to generate the formats? The directory you describe very likely shouldn't be there, and you can probably just delete it. Try renaming it to something else and see if that solves your problem. HTH Thomas On Jan 17, 2006, at 12:18 AM, David Wooten wrote:
I've rested a couple of days and found that, in fact, Gerben's distribution generates the new fmt files in ~/Library/texmf/web2c. Still, none of my attempts have produced changes in the allocated memory.
Hmm, interesting. I used "texconfig init". No, there seems to be no change if I rename it. Thanks, David On Jan 17, 2006, at 2:15 AM, Thomas A. Schmitz wrote:
No, that's not correct, it's not Gerben's distribution that does that. Which commands do you use to generate the formats? The directory you describe very likely shouldn't be there, and you can probably just delete it. Try renaming it to something else and see if that solves your problem.
HTH
Thomas
On Jan 17, 2006, at 12:18 AM, David Wooten wrote:
I've rested a couple of days and found that, in fact, Gerben's distribution generates the new fmt files in ~/Library/texmf/web2c. Still, none of my attempts have produced changes in the allocated memory.
1. Don't use texconfig for context, use texexec --make --all 2. If you use texonfig, the command is texconfig-sys, or you'll end up having inconsistent files for different users. Best Thomas On Jan 17, 2006, at 6:35 PM, David Wooten wrote:
Hmm, interesting. I used "texconfig init". No, there seems to be no change if I rename it.
Thanks, David
Thanks Thomas, I've still no luck, but I think things are quite a bit clearer now. I "stumbled" across an old fink installation, the existence of which hadn't crossed my mind for quite some time. That is now gone. I have also reinstalled my tex/context set-up via the i-Installer just to clear the air. Here is what I currently see: there are 3 cnf files that may be involved: /usr/local/teTeX/share/texmf.local/web2c/context.cnf Which declares: % This file is mostly the same as the one that ships with % texlive, but adapted for multiple trees as well as % development in a non tds tree (i.e. hh's machine). This % file is meant for context users. % % This file loads before texmf/web2c/texmf.cnf; beware : % the selfish auto parent can get in the way especially when % the binaries are located elsewhere. In case of multiple % trees, one can best set the following ones manually. I believe that Hans said this was just an example file (?), but it has the memory.context settings including: font_max.context = 2000 font_mem_size.context = 1000000 main_memory.context = 2000000 Second, there is: /usr/local/teTeX/share/texmf/web2c/texmf.cnf Which is referred to in the above context.cnf, and which declares: % 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. [Stuff skipped] % Earlier entries (in the same or another file) override later ones, and % an environment variable foo overrides any texmf.cnf definition of foo. This file also has memory settings, such as: % ConTeXt is a memory hog... main_memory.context = 2000000 % Keep total within 8000000 limit extra_mem_bot.context = 4000000 % Words of font info for TeX (total size of all TFM files, approximately). font_mem_size = 2000000 % Total number of fonts. Must be >= 50 and <= 2000 (without tex.ch changes). font_max = 2000 And finally, /usr/local/teTeX/texmf.cnf Which is identified by kpsewhich texmf.cnf and which has no memory settings. I have tried changing the memory settings in each file (that means adding them to /usr/local/teTeX/texmf.cnf), raising the main_memory to 3000000 and the font_mem_size likewise to 3000000, and then running texexec --make --all. No change in the resulting readout: Here is how much of TeX's memory you used: 4681 strings out of 60921 74346 string characters out of 610603 4609149 words of memory out of 7114466 40727 multiletter control sequences out of 10000+50000 1927616 words of font info for 332 fonts, out of 2000000 for 2000 232 hyphenation exceptions out of 5000 62i,21n,81p,1198b,1755s stack positions out of 5000i,500n,6000p, 200000b,40000s PDF statistics: 170 PDF objects out of 300000 0 named destinations out of 131072 109061 words of extra memory for PDF output out of 128383 At this point I would very much like to be embarrassed by a simple, too-obvious pointer ;) Thanks very much, David On Jan 17, 2006, at 10:31 AM, Thomas A. Schmitz wrote:
1. Don't use texconfig for context, use texexec --make --all
2. If you use texonfig, the command is texconfig-sys, or you'll end up having inconsistent files for different users.
Best
Thomas
On Jan 17, 2006, at 6:35 PM, David Wooten wrote:
Hmm, interesting. I used "texconfig init". No, there seems to be no change if I rename it.
Thanks, David
David Wooten wrote:
At this point I would very much like to be embarrassed by a simple, too-obvious pointer ;)
how about generating a plain tex format using pdfetex --ini and enabling the kpse debug options? Then you can see what is used: pdfetex --kpathsea-debug=65535 --ini --progname=context plain Hans
Hans, is this what you're referring to? kdebug:hash_lookup(main_memory.context) => 3000000 % Keep total within 8000000 limit 3000000 % Keep total within 8000000 limit kdebug:variable: main_memory = 3000000 % Keep total within 8000000 limit kdebug:hash_lookup(font_mem_size.context) => 3000000 kdebug:variable: font_mem_size = 3000000 kdebug:hash_lookup(font_max.context) => (nil) kdebug:hash_lookup(font_max) => 2000 kdebug:variable: font_max = 2000 If so, the increases I've made are evident here. . . but not when I run texexec? On Jan 17, 2006, at 2:36 PM, Hans Hagen wrote:
David Wooten wrote:
At this point I would very much like to be embarrassed by a simple, too-obvious pointer ;)
how about generating a plain tex format using
pdfetex --ini
and enabling the kpse debug options? Then you can see what is used:
pdfetex --kpathsea-debug=65535 --ini --progname=context plain
Hans
Any more clues, patient ones?: texexec --pdf file.tex Here is how much of TeX's memory you used: 1449181 words of font info for 262 fonts, out of 2000000 for 2000 pdfetex --kpathsea-debug=65535 --ini --progname=context plain kdebug:hash_lookup(font_mem_size.context) => 3000000 kdebug:variable: font_mem_size = 3000000 Many thanks, David
David Wooten wrote:
Any more clues, patient ones?:
texexec --pdf file.tex
How about: texexec --passon=-recorder --pdf file.tex That will give you a file named file.fls listing all used files including their path, and then you can easily check which fmt file is actually used. Cheers, taco
Oh, d**mn. I just had a brainwave. From Web2c's tex.ch comes the following: @!inf_font_mem_size = 20000; @!sup_font_mem_size = 2000000; So font_memory cannot be set larger than 2000000 and not smaller than 20000. Out-of-bounds sizes are silently folded back in line, but that happens after the variable expansion is complete. That's why the debug switch still reports the values from texmf.cnf. So the actual problem here is the web2c runtime code inside TeX. You can recompile with a larger sup_font_mem_size, I guess. But otherwise you are out of luck. Cheers, Taco David Wooten wrote:
Any more clues, patient ones?:
texexec --pdf file.tex
Here is how much of TeX's memory you used: 1449181 words of font info for 262 fonts, out of 2000000 for 2000
pdfetex --kpathsea-debug=65535 --ini --progname=context plain
kdebug:hash_lookup(font_mem_size.context) => 3000000 kdebug:variable: font_mem_size = 3000000
Many thanks, David _______________________________________________ ntg-context mailing list ntg-context@ntg.nl http://www.ntg.nl/mailman/listinfo/ntg-context
Taco Hoekwater wrote:
Oh, d**mn. I just had a brainwave.
From Web2c's tex.ch comes the following:
@!inf_font_mem_size = 20000; @!sup_font_mem_size = 2000000;
So font_memory cannot be set larger than 2000000 and not smaller than 20000. Out-of-bounds sizes are silently folded back in line, but that happens after the variable expansion is complete. That's why the debug switch still reports the values from texmf.cnf.
So the actual problem here is the web2c runtime code inside TeX.
You can recompile with a larger sup_font_mem_size, I guess. But otherwise you are out of luck.
so, is it possible to increase the upper limit in the code? hz seems to need it for the moment a way out is to limit hz to the regular font shape, i.e. don't use it for bold, italic, slanted etc another option is to limit the number of instances (steps) Hans ----------------------------------------------------------------- Hans Hagen | PRAGMA ADE Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: 038 477 53 69 | fax: 038 477 53 74 | www.pragma-ade.com | www.pragma-pod.nl -----------------------------------------------------------------
Hans Hagen wrote:
Taco Hoekwater wrote:
Oh, d**mn. I just had a brainwave.
From Web2c's tex.ch comes the following:
@!inf_font_mem_size = 20000; @!sup_font_mem_size = 2000000;
So font_memory cannot be set larger than 2000000 and not smaller than 20000. Out-of-bounds sizes are silently folded back in line, but that happens after the variable expansion is complete. That's why the debug switch still reports the values from texmf.cnf.
So the actual problem here is the web2c runtime code inside TeX.
You can recompile with a larger sup_font_mem_size, I guess. But otherwise you are out of luck.
so, is it possible to increase the upper limit in the code?
I guess so. I don't see any special reason for 2mln upper limit. font mem pointers are integers with rather straightforward usage, AFAIK. (CC to the tex-k list) Cheers, taco
Well, it is certainly very good to know that I'm not crazy (this time round). Thanks very much for tracking this down. Hans' suggestion works very well---the typescript file is pretty big for this font family, so there were quite a few fonts set up with hz that have no need for them. Many thanks, and if the chance arises (perhaps in San Francisco), many beers. David On Jan 19, 2006, at 1:44 PM, Hans Hagen wrote:
Taco Hoekwater wrote:
Oh, d**mn. I just had a brainwave.
From Web2c's tex.ch comes the following:
@!inf_font_mem_size = 20000; @!sup_font_mem_size = 2000000;
So font_memory cannot be set larger than 2000000 and not smaller than 20000. Out-of-bounds sizes are silently folded back in line, but that happens after the variable expansion is complete. That's why the debug switch still reports the values from texmf.cnf.
So the actual problem here is the web2c runtime code inside TeX.
You can recompile with a larger sup_font_mem_size, I guess. But otherwise you are out of luck.
so, is it possible to increase the upper limit in the code? hz seems to need it for the moment a way out is to limit hz to the regular font shape, i.e. don't use it for bold, italic, slanted etc another option is to limit the number of instances (steps) Hans
participants (4)
-
David Wooten
-
Hans Hagen
-
Taco Hoekwater
-
Thomas A. Schmitz