Bug#398354: tetex-bin: pdflatex segfaults
Gabor Gombas <gombasg@sztaki.hu> wrote:
> On Mon, Nov 13, 2006 at 07:20:27PM +0100, Hilmar Preusse wrote:
>
>> svn-book.pdf is the SVN book. Do I need a special pdf file to cause
>> the fault? Which options did you use?
>
> AFAIR I used "pdfnup --nup 2x1 --outfile /tmp/t.pdf article.pdf". The
> article was from a not-yet-published journal; I think it can be
> downloaded from Springer so if you have a Springer subscription I can
> look up the link, but I fear I'm not allowed to forward the article.
My university probably has a subscription, it would be great if you
could provide the link.
>> Weird. pdfLaTeX when be called from the command line directly works,
>> right?
>
> Just typing 'pdflatex' and then pressing 'Ctrl+D' works.
Yes, of course - that will neither touch fonts nor PDF inclusion code.
You'd have to grab the intermediate tex file that pdfnup creates.
> I think 0.5.0-1 was installed from experimental. I did not test/check
> sid but a clean etch chroot does work without a problem.
Please test in sid, everything points at a problem with experimental
libpoppler. We know that tetex-bin doesn't compile against experimental
libpoppler without patches.
> On AMD64 with
> the same library versions (including libpoppler) there is no SIGSEGV,
> just typing 'pdflatex' works, but the same pdfnup command still gives
>
> /usr/bin/pdflatex: symbol lookup error: /usr/bin/pdflatex: undefined symbol: _ZN6PDFDocC1EP9GooStringS1_S1_
>
> immediately, even without valgrind.
>
> Feel free to reassign the bug if you think it is a fault of libpoppler.
I think it's the fault of your using experimental. Maybe it's also a
bug of libpoppler not to bump its soname in experimental. The poppler
people are aware that they will need to do this, anyway - whether this
already applies to the package in experimental I don't know. But I'm
not going to care about it now, since we decided to delay a planned
poppler transition to after etch.
Regards, Frank
--
Dr. Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)
Reply to: