[csw-maintainers] TeXLive packaging take almost 3 days
Dagobert Michelsen
dam at opencsw.org
Mon Nov 19 15:49:58 CET 2012
Hi Peter,
Am 19.11.2012 um 15:44 schrieb pfelecan at opencsw.org:
> I'm struggling to replace teTeX with TeXLive since July... Quite a big
> endeavor. Here are some statistics:
>
> 112180 files in the installation directory
>
> 87 packages
>
> I'm using the split as defined by Debian with some adaptations.
>
> The time taken for each package is summarized in the following table:
>
> |-------+-------------------------------|
> | hh:mm | what |
> |-------+-------------------------------|
> | 67:09 | total_time |
> |-------+-------------------------------|
> | 04:17 | build_time |
> | 00:20 | global_prototype |
> | 00:36 | libkpathsea6 |
> | 00:36 | libptexenc1 |
> | 00:30 | tetex |
> | 00:41 | texlive-base |
> | 00:44 | texlive-bibtex-extra |
> | 00:43 | texlive-binaries |
> | 07:02 | texlive-common |
> | 02:48 | texlive-dev |
> | 00:31 | texlive-doc-base |
> | 00:32 | texlive-doc-bg |
> | 00:31 | texlive-doc-cs+sk |
> | 00:32 | texlive-doc-de |
> | 00:36 | texlive-doc-en |
> | 00:33 | texlive-doc-es |
> | 00:31 | texlive-doc-fi |
> | 00:32 | texlive-doc-fr |
> | 00:31 | texlive-doc-it |
> | 00:31 | texlive-doc-ja |
> | 00:31 | texlive-doc-ko |
> | 00:31 | texlive-doc-mn |
> | 00:32 | texlive-doc-nl |
> | 00:31 | texlive-doc-pl |
> | 00:32 | texlive-doc-pt |
> | 00:31 | texlive-doc-ru |
> | 00:31 | texlive-doc-si |
> | 00:31 | texlive-doc-th |
> | 00:31 | texlive-doc-tr |
> | 00:31 | texlive-doc-uk |
> | 00:33 | texlive-doc-vi |
> | 00:32 | texlive-doc-zh |
> | 00:36 | texlive-extra-utils |
> | 00:35 | texlive-font-utils |
> | 01:33 | texlive-fonts-extra |
> | 00:46 | texlive-fonts-extra-doc |
> | 00:52 | texlive-fonts-recommended |
> | 01:33 | texlive-fonts-recommended-doc |
> | 00:33 | texlive-formats-extra |
> | 00:37 | texlive-games |
> | 00:36 | texlive-generic-extra |
> | 00:32 | texlive-generic-recommended |
> | 00:35 | texlive-humanities |
> | 00:34 | texlive-humanities-doc |
> | 00:33 | texlive-lang-african |
> | 00:33 | texlive-lang-arabic |
> | 00:30 | texlive-lang-croatian |
> | 00:36 | texlive-lang-cyrillic |
> | 00:33 | texlive-lang-czechslovak |
> | 00:41 | texlive-lang-finnish |
> | 00:33 | texlive-lang-french |
> | 00:32 | texlive-lang-german |
> | 00:41 | texlive-lang-greek |
> | 00:33 | texlive-lang-hebrew |
> | 00:31 | texlive-lang-hungarian |
> | 00:34 | texlive-lang-indic |
> | 00:32 | texlive-lang-italian |
> | 00:31 | texlive-lang-lithuanian |
> | 00:32 | texlive-lang-mongolian |
> | 00:36 | texlive-lang-polish |
> | 00:31 | texlive-lang-portuguese |
> | 00:31 | texlive-lang-spanish |
> | 00:31 | texlive-lang-swedish |
> | 00:32 | texlive-lang-tibetan |
> | 00:35 | texlive-lang-vietnamese |
> | 00:36 | texlive-latex-base |
> | 00:34 | texlive-latex-base-doc |
> | 02:54 | texlive-latex-extra |
> | 01:47 | texlive-latex-extra-doc |
> | 00:51 | texlive-latex-recommended |
> | 00:40 | texlive-latex-recommended-doc |
> | 00:33 | texlive-latex3 |
> | 00:33 | texlive-luatex |
> | 00:42 | texlive-math-extra |
> | 00:33 | texlive-metapost |
> | 00:32 | texlive-metapost-doc |
> | 00:35 | texlive-music |
> | 00:35 | texlive-omega |
> | 00:38 | texlive-pictures |
> | 00:36 | texlive-pictures-doc |
> | 00:35 | texlive-plain-extra |
> | 00:57 | texlive-pstricks |
> | 00:40 | texlive-pstricks-doc |
> | 00:43 | texlive-publishers |
> | 00:40 | texlive-publishers-doc |
> | 00:38 | texlive-science |
> | 00:36 | texlive-science-doc |
> | 00:32 | texlive-xetex |
> | 00:32 | texlive-xindy |
> |-------+-------------------------------|
>
> Most of the time is spent in cswproto and patfilter utilities.
>
> For example: for libkpathsea6 the usage is the following:
>
> cswproto:
>
> real 1223.12
> user 1092.55
> sys 139.96
>
> of which 1077 by repeated calls to pkgproto
>
> pathfilter:
>
> real 2183.69
> user 2182.79
> sys 0.72
>
> IMHO, there is place for some optimization.
>
> If there is a kind soul to help on this he'll have my gratitude.
The numbers are… impressive. I have never optimized pathfilter as it usually
runs a second for average packages, so it probably leaves much room. Please use
DEBUG_PACKAGING=1 mgar repackage
to get the actual pathfilter invocation and grab the file it is applied to so I can
rerun and profile it. I can then happily try to speed it up.
Best regards
-- Dago
--
"You don't become great by trying to be great, you become great by wanting to do something,
and then doing it so hard that you become great in the process." - xkcd #896
More information about the maintainers
mailing list