Closed GoogleCodeExporter closed 9 years ago
This is a use case that I had never considered, so getting it right will
undoubtedly be
a lot of work.
There are two things that jump to mind, here:
1) it is possible in vim, emacs, and other tools, to set up different error
format
parsing (see :help errorformat), so the exact format of the error lines can
change with
relative impunity. We'd just create a special latex-makefile errorformat and
it would
understand the makefile's nicer output.
2) In order to make the above (or any solution, really) work properly, I will
have to
change the makefile to output line numbers.
So, that brings up a question: is the makefile already outputting filenames
correctly,
just not line numbers? If so, I may try tackling this. If not, it's probably
too big a
change for this project's lone maintainer :)
Original comment by shiblon
on 26 Aug 2009 at 1:02
I just realized that I made an attempt at educating you on 'errorformat', when
it's clear
you already understand that. I didn't note the name on the issue before
responding -
sorry about that. No patronizing intended.
Original comment by shiblon
on 26 Aug 2009 at 1:04
(Don't worry about it :))
Regarding 1), the errorformat might be the best solution, but I currently get no
filenames at all. This makes sense, since latex-makefile essentially greps the
output
for things that look like warnings, and there are none inline. So latex-makefile
would have to work out the filenames for this to work. (In theory, using
-file-line-error in pdflatex should print something better here, but I'm not
seeing
anything).
Regarding 2), yes, that true. That was issue 39 :)
FWIW, this isn't important to me anymore. I'm afraid to say I switched to
pdflatex
which latex-makefile doesn't support very well (a pity, because the other
alternatives are poor).
However, in experimenting with other solutions, I notice that rubber-info
provides an
amazing log parser, independently of rubber itself (that is, you can compile
with
latex-makefile, and still use rubber-info on the log file). It might be worth
reusing
it, if present.
Original comment by paul.biggar
on 26 Aug 2009 at 3:42
That's interesting about rubber.
So, you aren't the first to mention switching to pdflatex (see issue 6). Many
people
are going that route. That means I need to write a new makefile, which I just
jolly
well might go and do, because I hate the alternatives, too :-p
That said, my experience with it is not great. Can we take this discussion to
personal
email? I'd like to get some insights on
- why the switch to pdflatex instead of the dvi->ps->pdf approach
- what you intend to do to include pretty vector graphics in your document
(instead
of just jpg, etc.)
- how people are replacing the functionality of e.g., psfrag in pdflatex.
pdflatex has a simpler build process, so making a makefile for it shouldn't be
anywhere near as difficult as it was for the other, but I had very compelling
reasons
for sticking to dvi->ps->pdf, and psfrag was one of them. I obviously have a
lot of
passion about how difficult it is (but shouldn't be) to build LaTeX, so perhaps
we can
make something awesome for pdflatex, too.
Original comment by shiblon
on 26 Aug 2009 at 8:09
By the way, did you know you can do this:
make VERBOSE=1 myfile.pdf
And the makefile will print out the log as-is, with some things colorized? So,
when
using vim, you can still get your errorformat working fine just by turning on
the
VERBOSE setting.
That may not be what you want, in which case I can make a flag that only gets
verbose where vim needs it to do so. If you have some ideas about that, I'd be
happy
to discuss them.
Original comment by shiblon
on 28 Aug 2009 at 3:24
The verbose output here just confused vim. I've attached a log if its helpful.
I can
send you the .tex files on this (and other issues) if they're useful.
Original comment by paul.biggar
on 28 Aug 2009 at 11:08
Attachments:
Oh, of course it did. (slaps forehead). That setting is *really* verbose. I
should make a
verbose log setting that is separate.
But, it sounds like you have migrated away from this anyway. I'll fiddle with
it, but
recognize that it is no longer a priority for you to help debug stuff. Thanks
for your
input thus far, and good luck!
Original comment by shiblon
on 29 Aug 2009 at 3:20
Original comment by shiblon
on 19 Nov 2009 at 6:59
Original issue reported on code.google.com by
paul.biggar
on 19 Aug 2009 at 10:48