John Goerzen <[EMAIL PROTECTED]> wrote:
> severity 380247 normal
> thanks
>
> On Fri, Jul 28, 2006 at 06:54:16PM +0200, Frank Küster wrote:
>> Package: bacula-doc
>> Severity: serious
>> Tags: patch
>>
>> latex bacula.tex will fail, even when all needed depends are installed.
>
> I just built it again, and it works. What exactly are you seeing?
Ah, yes, sorry. It probably does not FTBFS, but only because the
commands that fail are prefixed with "-" in the Makefile, but I think
they shouldn't.
What I did was: I tried to find out which packages were needed to fix
the other FTBFS that was already reported, and therefore I ran "latex
bacula.tex", i.e. without -interaction=batchmode, and looked at the
errors. If you do the following change:
- -latex -interaction=batchmode bacula.tex
+ latex -interaction=nonstopmode bacula.tex
you can see them. Most of them are because the document uses the
underscore in non-math context without escaping it. There's also a
wrong nesting of environments at one place, and gfdl.tex uses preamble
commands after \begin{document}.
>> The files are full of further LaTeX problems: Use of the obsolete \bf
>> command instead of \bfseries (will cause trouble when fonts are changed,
>> and the letter kerning might already be amiss), declaration of a new
>> fontsize with hard-coded cmr which might also cause trouble, etc.pp.
>
> I would be happy to forward that on to upstream.
I fear just sending them a patch will not help in the long run, since
they are going to make the same mistakes again. Perhaps it's best if
they read l2tabuen.pdf, which nicely describes many errors that
old-fashioned or inexperienced LaTeX writers make. It's included in
most modern TeX distributions.
Will you contact them about this, or should I rather do it?
>> Note that the attached patch loads the non-free html.sty, because the
>> original document uses commands from it. I'm going to file a separate
>> bug about this.
>
> Unfortunately, that means I can't apply it.
Err, if I had been right that it needs html.sty, then it depends on it,
anyway - just ignoring the dependency wouldn't help. But in fact I was
wrong; it uses commands that also hyperref provides, but they are
undocumented - therefore I missed them.
> Until I see some actual evidence that there is an actual problem, I
> don't see what the problem is.
The Makefile succeeds only because upstream tells it to ignore any
errors, and there are many. I think the first one can be seen on the
page numbered 294, in the second paragraph below the heading
"performance issues": A variable make_xx_table is supposed to be
printed, but it should be written make\_xx\_table in the LaTeX source.
Without the backslashes, LaTeX is forced to math mode, interprets the
second underscore and prints the t as subscript and everything after the
first underscore as one bib multiplication of math symbols.
There are many more occurrences of this error - just find the places
where my patch escapes underscores.
On the following page there's a different error, the link is typeset as
http://www.postgresql.org/docs/faqs.FAQ.html#3.3, but it should end in
"#3.3". It doesn't work anyway, because it should be "#item3.3", but
also the internal representation is wrong.
I agree that this bug is not serious, since upstream's Makefile already
chooses to ignore the errors, and the resulting document is usefull,
even if buggy. Who would be the upstream person to contact about this?
Regards, Frank
--
Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)
