Re: [NTG-context] mapfiles and pdftex

At 08:53 AM 2/12/2003 +0100, you wrote:

Hartmut Henkel writes:

Hello Hartmut,

...

Don't know what is slow for your PC.

at least two seconds just for the map file.

...

I'm just looking into pdftex mapfile.c, about 2and there is in line 142 a for-loop for checking duplicate mapfile entries (tfm). This runs some 3000*0.5*3000 times for parsing your mapfile lines, always checking for some strcmp() == 0. You can speed this up if you put a line "if (*(e->tfm_name) != *buf) continue;"

But this also looks like O(n^2) to me. With 3000*3000 is some 9 million!

...

So there might be some corner for improvement. Using strcmp() the reverse way seems not to speed up things; how to write a more sophisticated inequality-checker?

Heiko stated that he used a hash table for this. This will speed up things dramatically.

but if i remember right he only looked at the first 8 chars (assuming berry) which is not that useful for texnansi-* files -) the best way (imo) is to let later entries overload previous ones, so not much checking is needed at all;

But this is getting slightly off topic, isn't it?

future versions of pdftex will solve this problem; another thing is reading map files after the first page is shiped out; that would permit us to postpone map file loading and speed things up a bit; currently, if you want to use a specific font on page 328 you must make sure that the associated map file is loaded before page 1 is finished! Hans ------------------------------------------------------------------------- Hans Hagen | PRAGMA ADE | pragma@wxs.nl Ridderstraat 27 | 8061 GH Hasselt | The Netherlands tel: +31 (0)38 477 53 69 | fax: +31 (0)38 477 53 74 | www.pragma-ade.com ------------------------------------------------------------------------- information: http://www.pragma-ade.com/roadmap.pdf documentation: http://www.pragma-ade.com/showcase.pdf -------------------------------------------------------------------------