Closed GoogleCodeExporter closed 9 years ago
Hey, a detailed bug with a minimal test case! That's awesome.
I'm not sure where the best place for a patch would be, either. The bibtex
dependency chain is the most esoteric in the Makefile, so I'll have to study
what's
really going on to figure out a better solution.
This use case is not all that uncommon. I'm surprised it hasn't come up before
now,
frankly. I'll take a look (but don't let that stop you from looking at it - I
may not get
to it for a week or more).
Original comment by shiblon
on 15 Mar 2010 at 3:41
Do me a favor and try out r0eb58b650c83. It *may* fix this (there was a super
subtle
dependency problem that might rear its ugly head here).
If it doesn't, that's fine, I'll still dig into it.
Original comment by shiblon
on 16 Mar 2010 at 2:13
Sadly that revision hasn't fixed this issue for me :(
Original comment by alexandr...@gtempaccount.com
on 16 Mar 2010 at 10:08
I've now been able to reproduce the issue. Working on it.
Original comment by shiblon
on 17 Mar 2010 at 6:51
So, it looks like the way to get this to build is to follow the following
procedure:
pdflatex ref
bibtex ref
pdflatex ref
bibtex ref
pdflatex ref
pdflatex ref
Is that right? Two invocations of bibtex required, with a pdflatex sandwiched
in
between, followed by two pdflatex invocations?
That's the only way I could get it to shut up. I'm currently looking at the
output
differences to see what changed so that I can (hopefully) automate this.
Original comment by shiblon
on 17 Mar 2010 at 6:59
Yes that's correct; that sequence works for both the test case I sent you and
my actual
document. I thought about hacking up the Makefile to just test for "undefined
references" warnings on the last run and if so rerun bibtex, but I don't think
that's
the right way to go about it?
Original comment by alexandr...@gtempaccount.com
on 17 Mar 2010 at 10:15
Yeah, the problem with blindly looking at "undefined references" is that
sometimes they
really are undefined :)
This is a bit tricky. I'm going to have to take a hard look at the steps taken
and see how
we can do this. It's not your typical invocation order.
Original comment by shiblon
on 18 Mar 2010 at 1:04
I just ran the sequence, saved all of the output files in different
directories, and did a
diff between runs. Nothing jumps out. There is no line in the output saying
"Run
again", there is no indication that anything may not have worked all the way.
Typically, packages emit something to the log when they need pdflatex to be run
again, but that is not the case here.
I'm not sure how to detect this, other than looking for apacann and forcing the
currently single bibtex run to be bibtex/pdflatex/bibtex (then letting the
other
machinery kick in as usual).
In fact, I think I'll try that, since nothing else seems like a likely solution.
Original comment by shiblon
on 18 Mar 2010 at 1:13
Fixed in r70ac667ecd0c. Basically, if I detect that bibtex needs to be run at
all, I run it
as before. Then, if I find that the style is apacann, I run pdflatex and
bibtex once more
before exiting the bbl target.
It works, and it doesn't appear to interfere with anything else, so I'm going
with it. Let
me know if it works for you.
If this process is needed for other kinds of bibstyles, I can easily add them.
Original comment by shiblon
on 18 Mar 2010 at 1:31
Awesome, thanks; that revision works perfectly on my test case and almost
perfectly
in my actual document (in which I also have citations within nomencl
nomenclatures -
but unlike before, I just need to repeat "make foo.pdf" once or twice and they
eventually get resolved).
Other annotated bibliography styles we should probably look for besides apacann
are:
apacannx (apacite package)
plain-annote
unsrt-annote
chicago-annote (chicago package)
chicagoa (chicago package)
IEEEannot
annotate
annotation
I think it's worth just checking if the style contains the string "annot", plus
the
special ones like chicagoa, apacann and apacannx, something like the attached
patch
(I've tested it against apacann, apacannx, plain-annote, chicago-annote,
chicagoa,
annotate and annotation, as well as plain (not annotated), and it seems to work
as
expected).
Original comment by alexandr...@gtempaccount.com
on 18 Mar 2010 at 11:05
Attachments:
Great patch. I edited it a tiny bit and incorporated it. See re218f0b2d2b3.
Hopefully this will also fix the problem you have with your actual document.
Original comment by shiblon
on 19 Mar 2010 at 5:21
This issue is already fixed, but I don't think it deserves a new one:
If you use the apacite bibliographystyle from the apacite packages the Makefile
doesn't work right. I attached a tiny patch. Hope you can use it.
Original comment by m...@diesuburbia.com
on 13 Dec 2010 at 6:30
Attachments:
Thanks! This is now incorporated into the tip (currently r442fefe278ee).
I haven't pushed a new build yet because there are massive changes going on in
overall project structure, but hopefully this will go in soon. Meanwhile, you
can pull the source down and run "build" to get a new Makefile if you want.
Original comment by shiblon
on 20 Dec 2010 at 6:07
Original issue reported on code.google.com by
alexandr...@gtempaccount.com
on 15 Mar 2010 at 4:24Attachments: