bugGNU Octave - Bugs: bug #48172, doc build fails with...

 
 

bug #48172: doc build fails with texi2dvi/texi2pdf errors on Ubuntu 14.04

Submitted by:  Muhali <muhali>
Submitted on:  Wed 08 Jun 2016 10:52:51 AM UTC  
 
Category:  Configuration and Build System Severity:  3 - Normal
Priority:  5 - Normal Item Group:  Build Failure
Status:  Fixed Assigned to:  None
Originator Name:  Open/Closed:  Closed
Release:  dev Operating System:  GNU/Linux

Add a New Comment (Rich Markup)
   

You are not logged in

Please log in, so followups can be emailed to you.

 

( Jump to the original submission)

Sat 02 Jul 2016 01:45:41 PM UTC, comment #58: 

Thanks Rik.  It didn't occur to me that the macro on the failing line would be a problem.  What I don't get is that some people said that there were just periodic and apparently random failures.  Maybe that was something else?

It is still a bug in Texinfo 5 though, because it works fine with 6.x.  But I agree it is OK to fix in Octave since the @nospell macro is not critical for us.

John W. Eaton <jwe>
Project Administrator
Sat 02 Jul 2016 05:01:36 AM UTC, comment #57: 

Confirmed to work here also, good job! I couldn't see that, and I still don't know what in the texinfo.tex changes caused this error to start happening, but I'll take it.

Mike Miller <mtmiller>
Project Administrator
Sat 02 Jul 2016 04:20:20 AM UTC, comment #56: 

On Ubuntu 14.04 (built with --without-OSMesa  option) no problem now
run until the end of tests with the following results:

Summary:

  PASS     14055
  FAIL         2
  XFAIL       29
  SKIPPED     54
Greetings

daniel chagas <danielch>
Sat 02 Jul 2016 03:14:30 AM UTC, comment #55: 

There was a problem.  I don't think Texinfo likes macros embedded within the @cindex macro.  I pushed a fix here (http://hg.savannah.gnu.org/hgweb/octave/rev/b602ce0b5acf).  This is ready for testing now.

Rik <rik5>
Project Administrator
Fri 01 Jul 2016 08:00:11 PM UTC, comment #54: 

Looks like the same error I get in comment #8 on fedora 22 and mint.

John Donoghue <lostbard>
Project Member
Fri 01 Jul 2016 06:26:15 PM UTC, comment #53: 

Mike, the error I get in doc/interpreter/octave.t2p/doc!interpreter!octave.t2d/pdf/build/octave.log with  make V=1  is:

[snip]
(/home/lachlan/src/tmp-octave/octave/.build/doc/interpreter/cp-idx.texi
(Concept Index) [969] [970]
Runaway definition?
->@initial {@} @entry {@nospell {Dulmage-Mendelsohn} decomposition}{5@ETC.
/home/lachlan/src/tmp-octave/octave/.build/doc/interpreter/cp-idx.texi:24: File
ended within @read.
<read 1>

@body ...ttoken @relax @else @read 1 to @nextline
                                                  @edef @act {@gdef @noexpan...

@next ->@let @next =@relax @body
                                 @let @next =@loopdoxxx @fi @next
@\printindex ... @let @thisline @nextline @repeat
                                                  @enddoublecolumns @fi @fi ...
l.24 @printindex cp

?
/home/lachlan/src/tmp-octave/octave/.build/doc/interpreter/cp-idx.texi:24: Argu
ment of @getfirsttokenx has an extra }.
<inserted text>
                @par
<to be read again>
                   }
@body ... @firsttoken {@getfirsttoken @nextline }}
                                                  @act @fi @thisline @ifeof ...

@next ->@let @next =@relax @body
                                 @let @next =@loopdoxxx @fi @next
@\printindex ... @let @thisline @nextline @repeat
                                                  @enddoublecolumns @fi @fi ...
l.24 @printindex cp

?
(Concept Index) [969] [970]
Runaway definition?
->@initial {@} @entry {@nospell {Dulmage-Mendelsohn} decomposition}{5@ETC.
/home/lachlan/src/tmp-octave/octave/.build/doc/interpreter/cp-idx.texi:24: File
ended within @read.
<read 1>

@body ...ttoken @relax @else @read 1 to @nextline
                                                  @edef @act {@gdef @noexpan...

@next ->@let @next =@relax @body
                                 @let @next =@loopdoxxx @fi @next
@\printindex ... @let @thisline @nextline @repeat
                                                  @enddoublecolumns @fi @fi ...
l.24 @printindex cp

?

I can send the whole .log file if that helps.

However, I agree this isn't a blocker for 4.2.0 since the docs are distributed pre-built.

Lachlan Andrew <lachlan>
Fri 01 Jul 2016 06:20:58 PM UTC, comment #52: 

Want to mark this as Won't Fix and close it?  If the future is 4.2, and this is going to affect only a small number of users, and there is no clear path to fixing this, then it doesn't seem worth it to leave it open.

Rik <rik5>
Project Administrator
Fri 01 Jul 2016 05:34:25 PM UTC, comment #51: 

I'm not sure I would classify this one as important (for the release). My justification is that it only affects users building from hg on older distributions. Users who build from the official 4.2.0 source distribution will have no problems.

And I still have no insight into whether this is a bug in the texinfo.tex macros file, or a bug in the interaction between it and Texinfo the software, or a bug in the Octave documentation (a typo or syntax error somewhere). The error message "etex exited with bad status" isn't particularly helpful, and I don't know how to debug the texi2dvi process or where to look for why it failed.

Mike Miller <mtmiller>
Project Administrator
Fri 01 Jul 2016 11:00:23 AM UTC, comment #50: 

Answering myself: no, as docs aren't built in mxe-octave.

Philip Nienhuis <philipnienhuis>
Project Member
Fri 01 Jul 2016 07:44:32 AM UTC, comment #49: 

Any use in comparing what is in the Linux host side and what is inside mxe-octave?
... as in the latter, buiding docs (at least for Windows) works fine w/o trickery a la comment #44.

Philip Nienhuis <philipnienhuis>
Project Member
Thu 30 Jun 2016 10:44:19 PM UTC, comment #48: 

For comment #44, doing the similar thing on my fedora box did not help

John Donoghue <lostbard>
Project Member
Tue 28 Jun 2016 07:50:15 PM UTC, comment #47: 

Latest dev still fails for me

John Donoghue <lostbard>
Project Member
Tue 28 Jun 2016 12:57:08 AM UTC, comment #46: 

I've just created a small demo project and I do not get an error when building a pdf of the trivial texinfo file I created.

By the way, there were some typos/syntax errors found in some Octave docstrings, maybe try building again?

Mike Miller <mtmiller>
Project Administrator
Fri 24 Jun 2016 04:08:54 PM UTC, comment #45: 

I can describe what I know about how this is supposed to work.

The upstream project GNU Texinfo contains the file of macros texinfo.tex that is maintained and improved within that project. It is a standalone file that can be included by any project when building their docs.

The GNU projects automake and gnulib both fetch texinfo.tex from the Texinfo project periodically and include the latest file verbatim in their repositories to be released with their code bases.

Octave uses automake to build, which looks first for a texinfo.tex file in Octave's build-aux directory, and if it's not found, it copies the texinfo.tex from its own repository (/usr/share/automake-VERSION) into Octave's build-aux and proceeds to build with that. So an automake-based project is never going to use the texinfo.tex file that is installed in /usr/share/texmf/tex/texinfo directory, the way I understand how the components work together.

Octave used to include its own hand-copied texinfo.tex file, but that was dropped recently because it had not been updated in years and was getting old and bit-rotten. IMHO that is not a maintainable solution.

No one has yet volunteered to pose a question on the texinfo project mailing list, which I think is the way to go forward on this.

I think it would also be reasonable for someone to create a small demo project that uses gnulib master to remove Octave from the equation and demonstrate that this error is a bug with one of those two projects.

Mike Miller <mtmiller>
Project Administrator
Fri 24 Jun 2016 03:43:47 PM UTC, comment #44: 

Indeed, as Lachlan suggested, just replacing the symlink in

<src dir>/build-aux/t@texinfo.tex

by one pointing to

/usr/share/texmf/tex/texinfo/texinfo.tex

makes for a smooth build of the docs.

BTW I saw these:

:
  GEN      doc/interpreter/splinefit6.txt
  MAKEINFO ../dev_VE/doc/interpreter/octave.info
  GEN      doc/interpreter/doc-cache
octave_doc_imkoq1:19529: misplaced {
octave_doc_imkoq1:19529: misplaced }
:

I tried to find where it came from but apparently octave_doc_imkoq1 is a scratch file that got deleted at the end of the build.

Philip Nienhuis <philipnienhuis>
Project Member
Fri 24 Jun 2016 11:33:02 AM UTC, comment #43: 

Just found this bug report.
I'm on Mageia-5 (x86_64) and have the very same issues since a few weeks but didn't bother yet. But now it I get also bitten when cross-building for Windows and I suspect this bug might be related.

Restarting the build (Linux) used to help. Running with just make (rather than make -j5) makes no difference. I'll try Avinoam's magic (comment #7) later today.

Something I observed:
octave.pdf is made long before the bulk of the docs. Later on, it seems to be built again but texi2dvi fails then. Doing a "touch doc/interpreter/octave.pdf" makes the build go on and finalize and ./run-octave& runs fine (and the docs seems to work).

I don't know if it is related, but since a few days (= before JWE's doc patch for core functions), after cross-building and installing in Windows, Octave complains that it cannot find ".../etc/macros.texi" when doing .e.g, "help more". And this error precludes installing OF packages.
mxe-octave was configured with "--disable-docs".

NB 1:
It could be that something is amiss with my system, but I doubt that it is related: when building Octave with JOBS=5 my Linux box hangs solid upon building/generating one of the .texi files (I think ~the 15th). "make -j1" avoids it.
I don't know which .texi as the screen saver had kicked in and I had to reboot "hard" so the status of the build was lost to me.
This started to happen after a glibc/kernel update. Running on the previous kernel version seems make it a little more stable.

NB 2:
IIRC a while (months) ago, Mike or Rik confirmed (in the maintainers ML?) that the doc build system sometimes needed restarting make one or more times.
Rik confirmed (also some months ago in a bug report) that building docs in mxe-octave worked for him with just one "job" (make -j1).

Philip Nienhuis <philipnienhuis>
Project Member
Thu 23 Jun 2016 12:37:52 PM UTC, comment #42: 

This may be a silly question, but isn't texinfo.tex part of texinfo?

If so, it isn't surprising that a file from texinfo 6.1 doesn't work with texinfo with a different major number.

Why do we need to have our own copy?  Is it just to remove one unknown of the environment?  If so, then when the configure process detects that the major number of the texinfo version doesn't match the one texinfo.tex came from, it could try to copy from /usr/share/texmf/tex/texinfo/texinfo.tex if that exists.  It solves the problem on my Linux Mint Debian Edition system.

Lachlan Andrew <lachlan>
Sat 18 Jun 2016 08:54:29 PM UTC, comment #41: 

Not that I expected it to, but the latest release of texinfo.tex (rev 7209) doesnt work either, so it hasnt accidentally been fixed in later versions.

John Donoghue <lostbard>
Project Member
Fri 17 Jun 2016 08:27:18 PM UTC, comment #40: 

Confirmed in my Ubuntu 14.04 VM, building the dvi and pdf fail with texinfo.tex from revision 7189, with or without the ^L characters that were added in that revision. Thank you for taking the time to investigate and bisect this.

So some change made to texinfo.tex in rev 7189 is not working with our doc build, with GNU Texinfo 5.2 and TeX Live 2013.20140215, but works fine with GNU Texinfo 6.1 and TeX Live 2016.20160523.

This commit was made on 2016-05-28, so still very fresh. Does anyone who understands Texinfo and/or TeX better than I do volunteer to discuss this on one of the texinfo mailing lists?

Mike Miller <mtmiller>
Project Administrator
Fri 17 Jun 2016 05:21:03 PM UTC, comment #39: 

I downloaded some of the revs between 7058 and 7190 - the first one to fail is http://svn.savannah.gnu.org/viewvc?view=rev&root=texinfo&revision=7189

The prev version (7187) worked ok.

 

Diff between those two versions is http://svn.savannah.gnu.org/viewvc/trunk/doc/texinfo.tex?root=texinfo&r1=7187&r2=7189

John Donoghue <lostbard>
Project Member
Thu 16 Jun 2016 11:43:00 PM UTC, comment #38: 

To be clear, this is the file "doc/texinfo.tex" in the GNU texinfo project repository, which is periodically copied into gnulib/build-aux/texinfo.tex, and is used as an input to build our documentation with whatever version of makeinfo/texi2dvi/texi2pdf you have installed on your build system.

Mike Miller <mtmiller>
Project Administrator
Thu 16 Jun 2016 11:40:15 PM UTC, comment #37: 

So by my inspection of the svn/git/hg logs of all projects involved, that means that this revision of upstream Texinfo works for you

http://svn.savannah.gnu.org/viewvc?view=rev&root=texinfo&revision=7058

This revision corresponds to the previous gnulib-hg that we were pinned to since the end of March 2016.

The latest gnulib-hg that I updated corresponds to this upstream Texinfo svn revision:

http://svn.savannah.gnu.org/viewvc?view=rev&root=texinfo&revision=7190

There were about a dozen changes to texinfo.tex in the Texinfo project between svn revision 7058 and 7190. It may be a little manual intensive but shouldn't be too difficult to bisect which revision it stops working on. And maybe that will give a clue as to what tool needs to be updated or what change was made that isn't backwards compatible.

Mike Miller <mtmiller>
Project Administrator
Thu 16 Jun 2016 08:28:06 PM UTC, comment #36: 

So after a few recompile attempts.

1. commenting out and/or removing the ctrl-L lines did not fix the issue in fedora

2. copying the old texinfo.tex from a checkout before that commit and using it on the current sources does work.

3. using the textinfo.tex from the current head of http://hg.octave.org/gnulib does not work (although I have not tried it with the ^L removed)

John Donoghue <lostbard>
Project Member
Wed 15 Jun 2016 06:23:48 PM UTC, comment #35: 

Old version had ^L characters but they were prefixed with comment characters. New revision has ^L immediately after a newline. Does deleting them do anything for you? If this is the problem, should be easy to see if upstream Texinfo has done anything about this, and if not report it to them.

Mike Miller <mtmiller>
Project Administrator
Wed 15 Jun 2016 06:18:05 PM UTC, comment #34: 

Diff between the old and new version of texinfo from different checkouts.
On my system there are some ^L - see line 35 for first in diff, that seem some somewhat weird to be used.

(file #37493)

John Donoghue <lostbard>
Project Member
Wed 15 Jun 2016 06:05:06 PM UTC, comment #33: 

Ok, good confirmation, so the gnulib update includes some new Texinfo things that don't work on Ubuntu 14.04 for some reason.

So next question, does copying the texinfo.tex file from an earlier revision of gnulib-hg fix this problem for you, when building on the default tip? You should first delete build-aux/texinfo.tex because it is normally a symlink into gnulib-hg, then drop the file into build-aux and build. That will definitively show that this file is the problem.

Someone could also look at upstream Texinfo commit log to see if there is anything suspicious, anything that may have broken with older TeX versions, something like that. Gnulib just pulls texinfo.tex from upstream Texinfo periodically, so we got a bulk update when I pushed that change.

Mike Miller <mtmiller>
Project Administrator
Wed 15 Jun 2016 06:00:11 PM UTC, comment #32: 

I guess no surprise, but on fedora box, result from bisect says:

changeset:   21820:96c56993e679
user:        Mike Miller <mtmiller@octave.org>
date:        Fri Jun 03 19:18:17 2016 -0700
summary:     maint: Update gnulib subrepo.

John Donoghue <lostbard>
Project Member
Wed 15 Jun 2016 01:37:06 AM UTC, comment #31: 

Tried to build revision 96c56993e679 on Ubuntu 14.04, failed with message :

 TEXI2DVI doc/interpreter/octave.dvi
/usr/bin/texi2dvi: etex exited with bad status, quitting.
make[2]: *** [doc/interpreter/octave.dvi] Error 1

daniel chagas <danielch>
Tue 14 Jun 2016 10:34:49 PM UTC, comment #30: 

I would not be surprised if 96c56993e679 is the first bad revision. It introduces a new texinfo.tex from a gnulib update. When I tested in an Ubuntu VM I tried copying the old texinfo.tex over, and I thought that was not the problem, but I may not have done it right.

Mike Miller <mtmiller>
Project Administrator
Tue 14 Jun 2016 09:18:50 PM UTC, comment #29: 

changeset 21810:1b5285c20e0e has some new warnings, but does not fail.

Avinoam Kalma <avinoam>
Project Member
Tue 14 Jun 2016 07:05:56 PM UTC, comment #28: 

starting bisection:

21820:96c56993e679 fails

21800:0eefa882bcc7  pass

Avinoam Kalma <avinoam>
Project Member
Tue 14 Jun 2016 12:20:21 PM UTC, comment #27: 

I started some bisects on the good vs bad build - slow going though as have to rebuild completely to verify it does in fact build/not build.

John Donoghue <lostbard>
Project Member
Tue 14 Jun 2016 12:07:53 PM UTC, comment #26: 

I'm getting the same on Ubuntu 14.04. I'll try and look at file history and commits to identify the cause. :)

TEXI2DVI doc/interpreter/octave.dvi
/usr/bin/texi2dvi: etex exited with bad status, quitting.
make[2]: *** [doc/interpreter/octave.dvi] Error 1
make[2]: Leaving directory `/home/nikola/git/octave'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/nikola/git/octave'
make: *** [all] Error 2

Nikola <nikola>
Mon 13 Jun 2016 09:25:50 AM UTC, comment #25: 

Just to point out that I am observing the same issue with openSUSE 13.2:

  TEXI2DVI doc/interpreter/octave.dvi
  MAKEINFO doc/interpreter/octave.html/.octave-html-stamp
/usr/bin/texi2dvi: etex exited with bad status, quitting.
Makefile:27041: recipe for target 'doc/interpreter/octave.dvi' failed
make[2]: *** [doc/interpreter/octave.dvi] Error 1

tex --version
TeX 3.1415926 (TeX Live 2013/TeX Live for SUSE Linux)

texi2dvi --version
texi2dvi (GNU Texinfo 4.13) 1.135

Guillaume <gyom>
Sun 12 Jun 2016 08:54:55 PM UTC, comment #24: 

Changing status to "need info" because the cause is not yet identified. It seems like it's a change in Octave since rewinding to a previous changeset fixes the build for some. So someone affected by this on Ubuntu 14.04 or Mint 17 may have to bisect to come up with the changeset where this started breaking.

Mike Miller <mtmiller>
Project Administrator
Sun 12 Jun 2016 08:01:10 PM UTC, comment #23: 

I did an explicit 'checkout' at that particular chageset and tried building it

John Donoghue <lostbard>
Project Member
Sun 12 Jun 2016 05:53:27 PM UTC, comment #22: 

John: Bisected? Or just the last changeset you remember building successfully with before some updates came in? Bisecting may be informative here.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 11:42:56 PM UTC, comment #21: 

And at:

> hg summary
parent: 21769:7496bd7265d9
doc: Document syntax for specifying color when using Tex interpreter (bug #47907).

it builds ok for me

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 09:46:59 PM UTC, comment #20: 

for comment #14 in mint 17:

  GEN      doc/interpreter/voronoi.eps
  GEN      doc/interpreter/triplot.eps

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 09:45:38 PM UTC, comment #19: 

On my mint17 box, I have texlive-full already installed and get the error

For comment #14, on my fedora box, looking back in the log I see:

 GEN      doc/interpreter/voronoi.eps
warning: print.m: epstool binary is not available.
Some output formats are not available.
warning: called from
    __print_parse_opts__ at line 382 column 9
    print at line 313 column 8
    geometryimages at line 55 column 5
warning: print.m: fig2dev binary is not available.
Some output formats are not available.
warning: called from
    __print_parse_opts__ at line 385 column 9
    print at line 313 column 8
    geometryimages at line 55 column 5
warning: print.m: pstoedit binary is not available.
Some output formats are not available.
warning: called from
    __print_parse_opts__ at line 388 column 9
    print at line 313 column 8
    geometryimages at line 55 column 5

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 09:32:05 PM UTC, comment #18: 

Muhali: do the latest comments on this bug report match your original report? I want to make sure we're not hijacking this report with a different unrelated symptom.

Can someone affected on Ubuntu 14.04 try installing more texlive packages, for example texlive-full?

This may have been caused by the recent update of the texinfo.tex macros file from gnulib. Previous updates of texinfo.tex have required more packages to be installed, e.g. texlive-fonts-recommended.

If installing more TeX packages helps resolve this, it would be very helpful if someone could identify exactly what packages need to be installed to provide the functions necessary to build.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 09:29:32 PM UTC, comment #17: 

I also get it on fedora 22

tex --version
TeX 3.14159265 (TeX Live 2014)

texi2dvi --version
texi2dvi (GNU Texinfo 5.2) 5381

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 07:38:29 PM UTC, comment #16: 

Ok, once I get past these osmesa and gnuplot issues, I can confirm the error messages with a Ubuntu 14.04 VM build, both texi2dvi and texi2pdf.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 07:28:52 PM UTC, comment #15: 

There seems to be another related bug, in that there is no dependency on scripts/plot/util/gnuplot_binary.m before the docs are built. I can get around this with

make scripts/plot/util/gnuplot_binary.m
make all

I'll report that separately.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 07:10:47 PM UTC, comment #14: 

Ok, so you're all on Ubuntu 14.04. Assuming you're building with OSMesa (i.e. not using the option --without-OSMesa), do you see the following messages in the build?

  GEN      doc/interpreter/voronoi.eps
error: __osmesa_print__: Depth and stencil doesn't match, are you sure you are using OSMesa >= 9.0?
error: unhandled execution exception -- eval failed
  GEN      doc/interpreter/triplot.eps
error: __osmesa_print__: Depth and stencil doesn't match, are you sure you are using OSMesa >= 9.0?
error: unhandled execution exception -- eval failed

This means the plot images are not created properly, even though the build is not failing at this point. Then when it tries to build the octave.dvi or octave.pdf, the included plots are not there.

And if you are building with --without-OSMesa, do the plots get created properly and the docs build successfully?

There is a known issue with the Ubuntu 14.04 libosmesa6 package, so there's no way to reliably build on 14.04 with OSMesa.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 05:01:34 PM UTC, comment #13: 

Ubuntu 14.04

$ tex --version
TeX 3.1415926 (TeX Live 2013/Debian)
kpathsea version 6.1.1

$ texi2dvi --version
texi2dvi (GNU Texinfo 5.2) 5381

Avinoam Kalma <avinoam>
Project Member
Sat 11 Jun 2016 04:37:09 PM UTC, comment #12: 

mint 17:

> tex --version
TeX 3.1415926 (TeX Live 2013/Debian)
kpathsea version 6.1.1

> texi2dvi --version
texi2dvi (GNU Texinfo 5.2) 5381

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 04:35:50 PM UTC, comment #11: 

ubuntu 14.04 : texinfo (5.2.0.dfsg.1-2)
daniel@newton:~$ tex --version
TeX 3.1415926 (TeX Live 2013/Debian)
kpathsea version 6.1.1
Copyright 2013 D.E. Knuth.

daniel chagas <danielch>
Sat 11 Jun 2016 04:25:08 PM UTC, comment #10: 

That's what the contents of cp-idx.texi look like in my build also, but no error.

Those affected by this, what versions of TeX and GNU Texinfo are you using?

I have TeX Live 20160320 and Texinfo 6.1.

Mike Miller <mtmiller>
Project Administrator
Sat 11 Jun 2016 04:15:47 PM UTC, comment #9: 

I'm on Ubuntu 14.04 64bits with mercurial sources
and can comfirm this, build stops with error on octave.dvi
greetings
daniel

daniel chagas <danielch>
Sat 11 Jun 2016 04:11:02 PM UTC, comment #8: 

with make V=1

(/home/jade/octave_hg/build/doc/interpreter/cp-idx.texi (Concept Index)
[969] [970]
Runaway definition?
->@initial {@} @entry {@nospell {Dulmage-Mendelsohn} decomposition}{5@ETC.
/home/jade/octave_hg/build/doc/interpreter/cp-idx.texi:24: File ended within @r
ead.
<read 1>

@body ...ttoken @relax @else @read 1 to @nextline
                                                  @edef @act {@gdef @noexpan...

@next ->@let @next =@relax @body
                                 @let @next =@loopdoxxx @fi @next
@\printindex ... @let @thisline @nextline @repeat
                                                  @enddoublecolumns @fi @fi ...
l.24 @printindex cp

?

build/doc/interpreter/cp-idx.texi is

@c DO NOT EDIT!  Generated automatically by munge-texi.pl.

@c Copyright (C) 1996-2015 John W. Eaton
@c
@c This file is part of Octave.
@c
@c Octave is free software; you can redistribute it and/or modify it
@c under the terms of the GNU General Public License as published by the
@c Free Software Foundation; either version 3 of the License, or (at
@c your option) any later version.
@c
@c Octave is distributed in the hope that it will be useful, but WITHOUT
@c ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
@c FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
@c for more details.
@c
@c You should have received a copy of the GNU General Public License
@c along with Octave; see the file COPYING.  If not, see
@c <http://www.gnu.org/licenses/>.

@node Concept Index
@unnumbered Concept Index

@printindex cp

John Donoghue <lostbard>
Project Member
Sat 11 Jun 2016 09:09:02 AM UTC, comment #7: 

To clarify comment #4:

I still get the error

  TEXI2DVI doc/interpreter/octave.dvi
/usr/bin/texi2dvi: etex exited with bad status, quitting.
make[1]: *** [doc/interpreter/octave.dvi] Error 1

The following black magic somehow solves it:

rm -rf ./doc/interpreter
make dvi
make
make

Avinoam Kalma <avinoam>
Project Member
Sat 11 Jun 2016 01:21:07 AM UTC, comment #6: 

I just started from scratch with rev 0331d3e832e3, ran ./bootstrap, made a clean build directory, ran my usual configure command, and was able to build completely with no errors. So still unable to reproduce here.

Mike Miller <mtmiller>
Project Administrator
Fri 10 Jun 2016 07:15:47 PM UTC, comment #5: 

I have the same problem on the dev version of octave in fedora 22 and Ubuntu 14.

Not sure went in happened, but it appears to work ok until I did a new ./bootstrap; mkdir build; cd build; ../configure; make

John Donoghue <lostbard>
Project Member
Fri 10 Jun 2016 10:04:52 AM UTC, comment #4: 

I have also encountered this problem, on Ubuntu 14.04.
somehow "make dvi" and "make" once or twice solved this.

Avinoam Kalma <avinoam>
Project Member
Wed 08 Jun 2016 09:33:27 PM UTC, comment #3: 

I am able to build completely, then `rm -rf ./doc/interpreter` and make again, and no errors for any of the doc targets here (building in a subdirectory of hg clone).

I am also able to `rm -rf ./doc/interpreter` and rebuild the docs again with MAKEINFO="$HOME/src/texinfo-4.13/makeinfo/makeinfo" which is a locally built version 4.13 (but still uses texi2dvi and texi2pdf version 6.1 in /usr/bin).

I don't see where this could be failing, TEXINPUTS is telling makeinfo that it can find preface.texi in the $(abs_top_builddir)/doc/interpreter directory.

Mike Miller <mtmiller>
Project Administrator
Wed 08 Jun 2016 08:37:26 PM UTC, comment #2: 

I meant octave.dvi, sorry. But either way, it gives identical errors for both.

In the Makefile I have

TEXINPUTS := $(PATH_SEPARATOR)$(abs_top_srcdir)/doc/interpreter$(PATH_SEPARATOR)$(PATH_SEPARATOR)$(abs_top_builddir)/doc/interpreter$(PATH_SEPARATOR)$(abs_top_builddir)/doc/.texmf-var//$(PATH_SEPARATOR)$(TEXINPUTS)$(PATH_SEPARATOR)
TEXMFCNF := $(PATH_SEPARATOR)$(abs_top_srcdir)/doc$(PATH_SEPARATOR)$(abs_top_builddir)/doc$(PATH_SEPARATOR)$(TEXMFCNF)$(PATH_SEPARATOR)

/usr/local/src/octave/build/doc/interpreter/preface.texi exists.

makeinfo (GNU texinfo) 4.13

Muhali <muhali>
Wed 08 Jun 2016 05:50:13 PM UTC, comment #1: 

Thanks for the bug report. I'm unable to reproduce here, and something doesn't seem quite right. The make error message you've quoted is about the "octave.pdf" target, but then you show a different error with the "octave.dvi" target. Is the same error occuring for you with both texi2dvi and texi2pdf?

What is the value of the TEXINPUTS variable in Makefile? Does /usr/local/src/octave/build/doc/interpreter/preface.texi exist?

What is makeinfo --version?

Mike Miller <mtmiller>
Project Administrator
Wed 08 Jun 2016 10:52:51 AM UTC, original submission:  

when trying to build octave from the current sources, from within a subdirectory 'build' and with docs enabled, the process stops with

  TEXI2PDF doc/interpreter/octave.pdf
/usr/bin/texi2dvi: pdfetex exited with bad status, quitting.
make[2]: *** [doc/interpreter/octave.pdf] Error 1
make[2]: Leaving directory `/usr/local/src/octave/build'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/usr/local/src/octave/build'
make: *** [all] Error 2

Digging a little bit into the building process it turns out that the culprit is when building the file octave.dvi:

TEXINPUTS="../build-aux:$TEXINPUTS"         MAKEINFO='/bin/bash /usr/local/src/octave/build-aux/missing makeinfo   -I doc/interpreter -I ../doc/interpreter'         texi2dvi --build-dir=doc/interpreter/octave.t2d -o doc/interpreter/octave.dvi  `test -f 'doc/interpreter/octave.texi' || echo '/usr/local/src/octave/build/../'`doc/interpreter/octave.texi

where I get

Please type another input file name: /usr/local/src/octave/build/../doc/interpreter/octave.texi:930: I can't find file `preface.texi'.

It works when building from within the original toplevel directory.

Muhali <muhali>

 

(Note: upload size limit is set to 16384 kB, after insertion of the required escape characters.)

Attach Files:
   
   
Comment:
   

Attached Files
file #37493:  difftexinfo.diff added by lostbard (18KiB - application/octet-stream)

 

Depends on the following items: None found

Items that depend on this one: None found

 

Carbon-Copy List
  • -email is unavailable- added by jwe (Posted a comment)
  • -email is unavailable- added by rik5 (Updated the item)
  • -email is unavailable- added by philipnienhuis
  • -email is unavailable- added by lachlan (Posted a comment)
  • -email is unavailable- added by nikola (Posted a comment)
  • -email is unavailable- added by gyom (Posted a comment)
  • -email is unavailable- added by danielch (Posted a comment)
  • -email is unavailable- added by lostbard (Posted a comment)
  • -email is unavailable- added by avinoam (Posted a comment)
  • -email is unavailable- added by mtmiller (Posted a comment)
  • -email is unavailable- added by muhali (Submitted the item)
  •  

    Do you think this task is very important?
    If so, you can add your encouragement to it.
    This task has 0 encouragements so far.

    Only project members can vote.

     

     

     

    Follow 13 latest changes.

    Date Changed by Updated Field Previous Value => Replaced by
    2016-07-02 mtmiller StatusReady For Test => Fixed
        Open/ClosedOpen => Closed
    2016-07-02 rik5 StatusConfirmed => Ready For Test
    2016-07-01 mtmiller Severity4 - Important => 3 - Normal
    2016-07-01 rik5 Severity3 - Normal => 4 - Important
    2016-06-30 philipnienhuis Carbon-Copy- => Added philipnienhuis
    2016-06-17 mtmiller StatusNeed Info => Confirmed
    2016-06-15 lostbard Attached File- => Added difftexinfo.diff, #37493
    2016-06-12 mtmiller StatusConfirmed => Need Info
    2016-06-11 mtmiller StatusWorks For Me => Confirmed
        Summarycannot build from within custom build directory => doc build fails with texi2dvi/texi2pdf errors on Ubuntu 14.04
    2016-06-08 mtmiller Item GroupNone => Build Failure
        StatusNone => Works For Me

    Back to the top


    Powered by Savane 3.5