Re: jadetex upload
Norbert Preining <preining@logic.at> wrote:
> On Mit, 02 Mai 2007, Frank Küster wrote:
>> > * remove the .ini hack, a wrongly set jobname/engine interferes with
>> > the search path ordering
>> > * The above together with depending on texlive 2007 fixes the interference
>> > with texlive-lang-cyrillic (Closes: #419026, #419020) [np]
>>
>> Can you explain to me how the search path is affected by the ini hack?
>> As I wrote earlier (I think), I found this, too, but didn't understand
>> it...
>
> The running "engine/format" combination was set to
> pdftex/etex/pdfetex
> whatever thus searching everywhere, while with
> &latex
> it uses the latex search pathes.
Hm, but fmtutil always uses "-progname=$format"[1], so shouldn't it be
jadetex all the time, no matter what the engine is?
According to "3.5.2.2 Determining the memory dump to use" in the web2c
manual, &DUMP takes precedence of the link name or -progname setting.
But that should only switch from jadetex to latex, and I get:
$ kpsewhich --progname=latex fonttext.cfg
/usr/share/texmf-texlive/tex/latex/base/fonttext.cfg
$ kpsewhich --progname=jadetex fonttext.cfg
/usr/share/texmf-texlive/tex/latex/base/fonttext.cfg
$ kpsewhich --progname=pdftex fonttext.cfg
/usr/share/texmf-texlive/tex/cslatex/base/fonttext.cfg
$
Therefore, even if preloading &latex switches to latex from jadetex, it
shouldn't affect the problem.
Actually, I think the cause for that bug is something different: just a
forgotten invocation of update-texmf. I did the following:
- log into a clean sid pbuilder chroot
- install tetex-extra (jadetex depends on it, and it brings
tl-lang-czechslovak and tl-lang-cyrillic
- install jadetex, it will fail
- run "dpkg --configure -a", it will fail
- run "update-texmf":
# update-texmf
Replacing config file /etc/texmf/texmf.cnf with new version
#
- run "dpkg --configure -a", it will *succeed*
The reason why I could reproduce the bug with my "nojadetex" format was
simply that no TEXINPUTS.nojadetex were defined...
However, of course I do think that dropping the ini hack is a good thing
if we are able to do that.
Regards, Frank
[1] except for metafun, and for finding the pool file
--
Dr. Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)
Reply to: