increasing max_strings
Hi, because of exceeded TeX capacity, I increased the following limit in the source: svn diff Index: source/src/texk/web2c/tex.ch =================================================================== --- source/src/texk/web2c/tex.ch (Revision 241) +++ source/src/texk/web2c/tex.ch (Arbeitskopie) @@ -200,7 +200,7 @@ @y @d file_name_size == maxint @d ssup_error_line = 255 -@d ssup_max_strings == 262143 +@d ssup_max_strings == 500000 {Larger values than 65536 cause the arrays consume much more memory.} @d ssup_trie_opcode == 65535 @d ssup_trie_size == @"3FFFFF Does anybody know offhand whether I can expect bad effects? After all, why would the ssup-setting not be the maximum possible value? And if there was a reason to set it that way, what other things would I have to change? I ask because the job that was now able to compile produced a corrupt PDF. There can be a number of other reasons for that, but since the run is quite slow, I'd want to have this ruled out if possible. Thanks, -- David Kastrup, Kriemhildstr. 15, 44793 Bochum
David Kastrup wrote:
Does anybody know offhand whether I can expect bad effects? After
I vaguely recall having problems when I tried to raise the value, but I can't remember what was wrong, and I am not even sure anymore if I crossed the 16-bit boundary, sorry. The value 262143 is probably only a legacy of 36-bit words on the PDP-10, but ... Best wishes, Taco
Taco Hoekwater
David Kastrup wrote:
Does anybody know offhand whether I can expect bad effects? After
I vaguely recall having problems when I tried to raise the value, but I can't remember what was wrong, and I am not even sure anymore if I crossed the 16-bit boundary, sorry.
The value 262143 is probably only a legacy of 36-bit words on the PDP-10, but ...
I did some sort of cursory code review of pdftex.web and found nothing obvious that would go wrong. I am still somewhat queasy about this. Perhaps Hans has some anecdotal evidence? I could imagine that the Pragma workflow will tend to favor PDFTeX compilations with large limits. -- David Kastrup, Kriemhildstr. 15, 44793 Bochum
participants (2)
-
David Kastrup
-
Taco Hoekwater