Fri 01 Sep 2017 11:42:34 AM UTC, comment #6:
V=1 isn't something specific to the rules for building documentation, it's a general feature of Makefiles generated by automake when using silent rules. So this problem applies to all failed targets. If you want to see the commands used, you need to use V=1.
OTOH, the TeX targets are a little different, because you don't even see the error messages. That's because they are written to stdout along with a lot of other log info instead going to stderr, and there's no easy way to just display the errors. So automake just hides all the output from these rules when in silent mode. Fixing this seems like more trouble than it is worth. This is a developer problem anyway. Typical users who build from a tarball shouldn't need to rebuild DVI files.
Closing as "won't fix".
|
Fri 01 Sep 2017 07:25:29 AM UTC, comment #5:
One simple thing that could be done to help the user is to supply the V=1 hint when there is a doc build error, i.e.:
|
Fri 01 Sep 2017 07:07:12 AM UTC, comment #4:
OK, it wasn't a total rebuild...
So, now I'm getting some good info:
Installing epsf.tex did the trick. Thanks. (Mint 18.2 seems to group things a little differently. Maybe some dependencies were missed.)
|
Fri 01 Sep 2017 06:28:46 AM UTC, comment #3:
That is what I've been trying, but could never get anywhere. Thanks for the make V=1 tip.
The settings for the comment characters is (via grep):
config.h:#define ENABLE_DOCS 1
config.log:#define ENABLE_DOCS 1
config.status:D["ENABLE_DOCS"]=" 1"
and in configure:
config.log:AMCOND_BUILD_DOCS_FALSE='#'
config.status:S["AMCOND_BUILD_DOCS_FALSE"]="#"
Well, I did a ./bootstrap ten minutes ago, so make is going to take a while. I'll write back (probably tomorrow).
|
Fri 01 Sep 2017 06:20:06 AM UTC, comment #2:
Sorry, that should be 'make V=1', to have the build commands echoed fully.
|
Fri 01 Sep 2017 06:18:39 AM UTC, comment #1:
You're not reading the makefile correctly, please look again.
There should be a commented-out rule for doc/interpreter/octave.dvi that you quoted, but if you continue searching there will be a rule that is not commented-out.
The reason for the error is not a makefile configuration error, it's simply an error exit status from the etex program. If you want to know what that is, you should probably try 'make V=0' to see the full build command, and then try to run the texi2dvi command manually and go from there.
|
Fri 01 Sep 2017 05:52:14 AM UTC, original submission:
I've run into a build problem:
It seems to manifest similar to
https://savannah.gnu.org/bugs/?48172
and discussion thread
http://octave.1599824.n4.nabble.com/Fwd-Re-make-failed-tt4678092.html#a4678102
I got stuck on the fact I had a Sundials IDAKLU issue:
, but after trying a half dozen variations on including and excluding OSMesa, Sundials and Java, I came to the conclusion that none of that seems to matter.
I can tell what is wrong. The octave.dvi file is NOT being generated because it is conditionally commented out of the Makefile:
and I can verify that is the case by observing in the Makefile.in file:
So, something is indicating that octave.dvi should not be created. Whether that is a missing TeX file or something, I will investigate. However, the issue is that because octave.dvi is conditionally excluded from the build, then too should the
TEXI2DVI doc/interpreter/octave.dvi
that attempts to utilize the octave.dvi be excluded.
|