Open GoogleCodeExporter opened 9 years ago
Just so you know this isn't a black hole, I'll look into this. It may not be
immediate, but I'll get to it :)
Things are always faster with patches, but I don't expect that; I only
encourage it
when people feel like they can manage to navigate this labyrinthine file.
Original comment by shiblon
on 26 May 2009 at 3:31
That's very strange behavior, given that there is a rule to build %.ind files
from %.idx
and %.tex files. I think what happened is this: in order to build a %.ind
file, the tex and
idx files have to have the same name.
...that's sort of an interesting issue to have, isn't it? I would benefit from
a very small
example of files that cause this issue being attached to this bug. I believe
we can get it
fixed, but I don't have time to muddle through making a test case.
Original comment by shiblon
on 19 Jun 2009 at 2:36
This is probably an edge case in terms of functionality. The issue is I have a
work
that was authored by several people. I want to index the contributions of each
author. My understanding is the LaTeX Way of doing this is to create a separate
index
of each author. The code example I provide demonstrates creating one of the
separate
indices.
I gather you would have to scan *tex files for \printindex[{index name}], to
build
the list of non-standard index names.
Original comment by dau...@gmail.com
on 19 Jun 2009 at 3:29
I can almost certainly pull it out of LaTeX's error log, which is what I do now
to create
index rules.
If you can come up with a very small test case (complete files that I can just
attempt to
build, that give the appropriate error), then I'll work on this. Just attach
them to the
bug.
Original comment by shiblon
on 19 Jun 2009 at 5:16
By the way, I'm still happy to look into this - I could just use some help with
it. In
particular, if you could download the latest version of the makefile, run it on
your test
case, and attach a tarball with all of the .log files, I'd be grateful.
Original comment by shiblon
on 19 Nov 2009 at 6:48
Issue 58 has been merged into this issue.
Original comment by shiblon
on 1 Dec 2009 at 3:45
Sorry, I got distracted. I can touch on this this week and give you a test...
Original comment by dau...@gmail.com
on 1 Dec 2009 at 4:15
No problem. I believe I have this fixed in r7f3173f3607e. Please download and
have a
go with it. If it works, I'll upload a new tarball.
Ironically, the hardest part was making sure that make "clean" worked properly,
refusing to delete indices that are hand-generated. I think I got it, though
(although
sometimes it just won't work, if the .fls file goes away after a build, for
example).
Original comment by shiblon
on 1 Dec 2009 at 5:31
The issue isn't fixed yet.
The line-numbers are to low.
Invocing makeindex manually after "make" is run delivers the right line-numbers.
Probably "makeindex" and "Latex" both have to be called once more after the
layout
has settled.
I'm working on an example, but this may take some time.
Original comment by peter.tr...@t-online.de
on 3 Dec 2009 at 7:10
Please elaborate on what you mean by "line numbers are too low." Which line
numbers,
and where is the output wrong? Does it affect the document, or the
informational
output of make?
Original comment by shib...@google.com
on 3 Dec 2009 at 7:12
Original comment by shiblon
on 3 Dec 2009 at 7:13
Sorry for beeing inaccurate.
I didn't mean linenumers, I meant the pagereferences delivered by the index (in
the dvi).
The "latex" command apparendly produces one or more idx-files every-time it is
run.
The structure (wraps etc.) of the dvi-document may still change after the first
run.
"makeindex" is probably run after the first run of "latex", but the pagenumbers
still
change after that and so the pagereferences point to an "old" page.
Original comment by peter.tr...@t-online.de
on 3 Dec 2009 at 8:25
Okay, I get it. That just means that latex needs to be run more than once, not
that
makeindex needs to be run again, doesn't it? I don't see anything in the
makeindex
output that would cause this, but I can see how running LaTeX again would be
necessary in these cases.
Original comment by shiblon
on 4 Dec 2009 at 1:42
If you could do me a favor, I'd really appreciate it. Build your output file
using the
makefile (complete with incorrect references), then tar up and attach all files
ending in
.fls or .log to this issue. Then I can find out whether there is a way to
detect this
condition and run latex again.
Original comment by shiblon
on 4 Dec 2009 at 1:46
I really need this information to proceed. Meanwhile, I'm planning on
releasing a new
version number. The .log files from your build (and actually the .aux files,
too) will
really help me debug this.
Original comment by shiblon
on 10 Dec 2009 at 2:46
Downloaded the latest version. The file it carps about is in the same directory
as
the Makefile...
[bwilson@bwilson-laptop]~/Books/MacropediaTerradoma$ make all
NOTE: You may ignore warnings about the following files:
MacropediaTerradoma.d
Makefile:1809: MacropediaTerradoma.d: No such file or directory
= MacropediaTerradoma.tex --> MacropediaTerradoma.d MacropediaTerradoma.dvi (1)
=
Missing argument. Try `kpsewhich --help' for more information.
make: *** No rule to make target `BorisHelfstein.ind', needed by
`MacropediaTerradoma.d'. Stop.
Original comment by dau...@gmail.com
on 10 Dec 2009 at 11:13
Attachments:
Aha! Which latex distribution are you using? It looks like the problem is
with my
invocation of kpsewhich.
I'm looking at your log files, now. Thanks for sending them.
Were there other files ending in .log in your directory? For example, files
ending in
.1.log, .2.log, etc.?
Original comment by shiblon
on 10 Dec 2009 at 3:58
OK. The log you sent me has a "Rerun to get cross references right" warning in
it.
The Makefile should have noticed this and run again.
But, it might not have done so because it caps the number of times it will run
by
about 5. Are you, by any chance, bumping up against this limit? I can
increase it
easily enough.
In fact, so can you, just to test. There is a line in the file that says "for
i in 2 3 4 5;
do" If you add more numbers on the end, like "for i in 2 3 4 5 6 7; do" you
can get it
to run LaTeX a few more times if it finds it necessary.
If that fixes your incorrect page number issues, I'd love to know about it.
Meanwhile, if you could also include the output of kpsewhich --help in this
issue
report, I can tackle your other problem.
Original comment by shiblon
on 10 Dec 2009 at 4:03
I am pretty sure that kpsewhich is a red herring here. Please try increasing
the number
of invocations of latex as described above and ping me with the results.
Original comment by shiblon
on 9 Mar 2010 at 8:49
Original issue reported on code.google.com by
dau...@gmail.com
on 22 May 2009 at 1:11