Bug#320821: tetex-bin: Latex sometimes produces (useless) dvi files for broken tex files.
Hi Bas,
Bas Kloet <bas@luon.net> wrote:
>> You cannot stop this "document" from producing a dvi file, unless you
>> change the bowels of the TeX language, which will probably not even
>> happen if some day people in fact manage a complete rewrite of TeX, as
>> is currently planned.
>
> Ok, I can understand that. One strange thing I found after further
> investigation is that both 'pdftex' and 'pdflatex' fail on the
> previously mentioned TeX file, but that only pdflatex produces a
> (broken) pdf file. If whatever stops pdftex from creating a pdf file in
> this case could be implemented in pdflatex as well then at least part of
> the problem will be solved.
It is the other way round: latex loads the LaTeX format before your
file, and in this format various things are done, including page setup.
It is not something that is implemented in tex, but rather what is
missing in the plain format - no page setup - that results in no dvi/pdf
file being produced. Again, we would have to take away from LaTeX
exactly what makes that LaTeX is LaTeX to get the "wanted" behavior.
>> I am quite sure you can produce equally useless binaries with a
>> compiler, if you put enough effort in writing carefully designed
>> nonsense in the source code, and you wouldn't blame the compiler.
>
> It's not that carefully crafted at all, I just removed all useless cruft
> to pinpoint the problem. The document which created this problem had a
> broken "\begn{document}", followed by pages of text with one \newline in
> it. This produced the totally broken pdf file.
I suggest not to use scripts like rubber early in the development cycle,
but rather something that shows you the errors. This could be running
latex on the command line, or AUCTeX in Emacs. Both would have shown
you the error message "missing \begin{document}" even before you'd
removed "useless cruft" like \documentclass...
> I saw that you also CC'd this reaction to the 'rubber' bug I filed. I
> agree that the LaTeX part of the bug may be dubious, but rubber should
> really detect that pdflatex exited with an error and remove the pdf
> file.
What's the problem with having an unreadable file around, when you have
to go debugging, anyway? I'd rather not have rubber remove the output
file upon every error - some (actually most) TeX errors are all but
fatal, and you can very well work with the output file. And there's no
way for rubber to discriminate between them.
Regards, Frank
--
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer
Reply to: