help-octave
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: compiling octave fails on building flex


From: John Donoghue
Subject: Re: compiling octave fails on building flex
Date: Sun, 17 Dec 2017 08:44:01 -0500
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.5.0

On 12/16/2017 03:10 PM, Marshall Floyd wrote:
On 12/15/2017 10:55 AM, Dmitri A. Sergatskov wrote:


On Fri, Dec 15, 2017 at 10:26 AM, address@hidden <address@hidden> wrote:
> If you run 'make hg-octave-dist', it will attempt to create a updated octave dev tarball that then will be used when running make nsis-installer. If you don't run that, it will use the tarball that it downloaded the first time it compiled the dev octave files.

I have been having troubles building with 'make hg-octave-dist' AND 'make nsis-installer'. I've already reported that I got the make nsis-installer working on one machine. On my other machine I got 'make hg-octave-dist' to work and was able to run it with './run-octave', and see that the last few features I had wanted worked on the linux version. However, I can't get either machine to run the alternate 'make' command to work. I continually get errors on one for building the native, and on the other can't get the windows installer. I should have saved the linux version when I had it working, to move it to the machine where the windows installer worked, but now it won't compile it again for some reason. I feel like somehow the order of building affected things because running things again have different errors, and I think that was the difference between the two machines, or maybe which dependencies I installed (the various lists of apt-get install ... lists). I'll try again from the start with a clean repo, but I expect I'll run against the same errors so maybe someone already knows the causes and how to fix some of these.

 

The following commands result in the following errors:

$make clean

$./configure --enable-devel-tools --enable-octave=default --enable-qt5 --enable-64 --enable-windows-64 --disable-docs

$make hg-octave-dist -j6 <or without -jN>

...

running: libtoolize --install --copy

libtoolize:   error: 'build-aux/config.guess' is newer: use '--force' to overwrite

libtoolize:   error: 'build-aux/config.sub' is newer: use '--force' to overwrite

libtoolize:   error: 'build-aux/install-sh' is newer: use '--force' to overwrite

libtoolize: Consider adding '-I m4' to ACLOCAL_AMFLAGS in Makefile.am.

...

GEN      doc/interpreter/interpft.eps

warning: division by zero

warning: called from

    legend at line 655 column 14

    interpimages at line 43 column 5

warning: division by zero

warning: called from

    legend at line 831 column 20

    interpimages at line 43 column 5

...

GEN      doc/interpreter/splinefit1.eps

warning: division by zero

warning: called from

    legend at line 655 column 14

    splineimages at line 52 column 5

warning: division by zero

warning: called from

    legend at line 840 column 20

    splineimages at line 52 column 5

...

warning: division by zero

warning: called from

    legend at line 831 column 20

    splineimages at line 162 column 5

warning: division by zero

warning: called from

    legend at line 916 column 11

    splineimages at line 162 column 5

  TEXI2DVI doc/interpreter/octave.dvi

/usr/bin/texi2dvi: etex exited with bad status, quitting.

Makefile:28017: recipe for target 'doc/interpreter/octave.dvi' failed

make[3]: *** [doc/interpreter/octave.dvi] Error 1

...

 

------------------------------

On the other machine, it doesn't choke on octave.dvi (and had worked all the way through once, but now gives the following:

 

...

  TEXI2DVI doc/interpreter/octave.dvi

  DVIPS    doc/interpreter/octave.ps

  GEN      doc/interpreter/voronoi.pdf

warning: division by zero

warning: called from

    legend at line 655 column 14

    geometryimages at line 54 column 5

...

warning: division by zero

warning: called from

    legend at line 916 column 11

    geometryimages at line 54 column 5

Error: /undefined in -nan

Operand stack:

   --nostringval--

Execution stack:

   %interp_exit   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval--   --nostringval--   --nostringval--   false   1   %stopped_push   1999   1   3   %oparray_pop   1998   1   3   %oparray_pop   --nostringval--   1982   1   3   %oparray_pop   1868   1   3   %oparray_pop   --nostringval--   %errorexec_pop   .runexec2   --nostringval--   --nostringval--   --nostringval--   2   %stopped_push   --nostringval--

Dictionary stack:

   --dict:1215/1684(ro)(G)--   --dict:0/20(G)--   --dict:83/200(L)--   --dict:41/64(L)--

Current allocation mode is local

GPL Ghostscript 9.21: Unrecoverable error, exit code 1

  GEN      doc/interpreter/triplot.pdf

  GEN      doc/interpreter/griddata.pdf

...

<with more division by zero warnings and GPL Ghostscript errors and other files>

...

  MAKEINFO doc/interpreter/octave.html/.octave-html-stamp

doc/interpreter/poly.texi:790: warning: @image file `splinefit1' (for HTML) not found, using `splinefit1.jpg'

doc/interpreter/poly.texi:834: warning: @image file `splinefit2' (for HTML) not found, using `splinefit2.jpg'

doc/interpreter/poly.texi:871: warning: @image file `splinefit3' (for HTML) not found, using `splinefit3.jpg'

doc/interpreter/poly.texi:910: warning: @image file `splinefit4' (for HTML) not found, using `splinefit4.jpg'

doc/interpreter/poly.texi:954: warning: @image file `splinefit6' (for HTML) not found, using `splinefit6.jpg'

doc/interpreter/interp.texi:168: warning: @image file `interpderiv1' (for HTML) not found, using `interpderiv1.jpg'

doc/interpreter/interp.texi:174: warning: @image file `interpderiv2' (for HTML) not found, using `interpderiv2.jpg'

doc/interpreter/interp.texi:233: warning: @image file `interpft' (for HTML) not found, using `interpft.jpg'

doc/interpreter/geometry.texi:649: warning: @image file `voronoi' (for HTML) not found, using `voronoi.jpg'

  GEN      doc/interpreter/octave.html/voronoi.png

cp: cannot stat 'doc/interpreter/voronoi.png': No such file or directory

Makefile:28045: recipe for target 'doc/interpreter/octave.html/voronoi.png' failed

make[3]: *** [doc/interpreter/octave.html/voronoi.png] Error 1

...

 

-----------------

The division by zero- warnings didn't use to show up in my first failed attempts at make hg-octave-dist, so I'm not sure why they started showing up. The failure on octave.dvi is the critical failure that has always showed up on the one machine however. I note that there's an octave bug, bug #48172, that mentions the same error, but was supposedly fixed for an older version of Ubuntu. I try some of the same things there and they haven't worked (e.g. configuring with "--disable-docs", which doesn't seem to do anything, or running make again, or replacing the symlink to texinfo.tex). Looking at a log when  make V=1 is done, gives a different set of errors than in that bug report (I wish I had copied that before trying other builds, because now I don't have that log anymore).




​This looks like OpenGL/gl2ps issue to me -- the figures are not generated.

Try
​LIBGL_ALWAYS_SOFTWARE=1 make

may be also try to preload mesa libs (though I think the error is usually different)

LD_PRELOAD=/usr/lib64/libGLX_mesa.so.0 LIBGL_ALWAYS_SOFTWARE=1 make

​(path and name of the mesa lib may be different on your system)

​Dmitri.
--

​LIBGL_ALWAYS_SOFTWARE=1 make, didn't seem to make a difference. I've started to keep track of exactly what order of commands I do while building to avoid repeating or get intuition of what changes between and that's help narrow things down. After a while and looking at errors and getting other errors resolved, the only remaining errors appear to be those relating to errors in generating the images needed for the docs.

On my other machine though, I got the linux version compiled again, and ./octave-hg-repo/.build/run-octave opens octave and has the new bug fixes I was looking for. make nsis-installer -j4 -k after multiple runs fails with the following three errors:

[build]    nsis
[build]    of-general
[build]    gdb

Failed to build package nsis!
------------------------------------------------------------
scons: *** Path for option APPEND_LIBPATH does not exist: /home/marshall/mxe-octave/usr/lib32
File "/home/marshall/mxe-octave/tmp-nsis/nsis-2.50-src/SConstruct", line 177, in <module>
/home/marshall/mxe-octave/Makefile:851: recipe for target 'build-only-nsis' failed
make[1]: *** [build-only-nsis] Error 2
make[1]: Leaving directory '/home/marshall/mxe-octave'
real    0m2.981s
user    0m1.862s
sys    0m0.458s
------------------------------------------------------------
[log]      /home/marshall/mxe-octave/log/nsis

Makefile:851: recipe for target '/home/marshall/mxe-octave/installed-packages/nsis' failed
make: *** [/home/marshall/mxe-octave/installed-packages/nsis] Error 1

Failed to build package of-general!
------------------------------------------------------------
  File "/home/marshall/mxe-octave/tools/pkg-install.py", line 237, in configure_make
    raise Exception, "make failed during build - stopping install"
Exception: make failed during build - stopping install
/home/marshall/mxe-octave/Makefile:851: recipe for target 'build-only-of-general' failed
make[1]: *** [build-only-of-general] Error 1
make[1]: Leaving directory '/home/marshall/mxe-octave'
real    0m12.185s
user    0m10.556s
sys    0m0.802s
------------------------------------------------------------
[log]      /home/marshall/mxe-octave/log/of-general

Makefile:851: recipe for target '/home/marshall/mxe-octave/installed-packages/of-general' failed
make: *** [/home/marshall/mxe-octave/installed-packages/of-general] Error 1

Failed to build package gdb!
------------------------------------------------------------
Makefile:849: recipe for target 'all' failed
make[2]: *** [all] Error 2
make[2]: Leaving directory '/home/marshall/mxe-octave/tmp-gdb/gdb-8.0.1'
/home/marshall/mxe-octave/Makefile:851: recipe for target 'build-only-gdb' failed
make[1]: *** [build-only-gdb] Error 2
make[1]: Leaving directory '/home/marshall/mxe-octave'
real    8m22.099s
user    7m20.136s
sys    0m47.410s
------------------------------------------------------------
[log]      /home/marshall/mxe-octave/log/gdb

Makefile:851: recipe for target '/home/marshall/mxe-octave/installed-packages/gdb' failed
make: *** [/home/marshall/mxe-octave/installed-packages/gdb] Error 1
make: Target 'nsis-installer' not remade because of errors.



nsis fails I think because it tries to access the non-existent directory mxe-octave/usr/lib32

of-general fails according to the log because of , though there are other warnings of deprecated calls:

packfields.cc:54:39: error: conversion from ‘octave::symbol_table::scope*’ to non-scalar type ‘octave::symbol_table::scope’ requested
        = interp.require_current_scope ("packfields");
          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~
...
unpackfields.cc:60:39: error: conversion from ‘octave::symbol_table::scope*’ to non-scalar type ‘octave::symbol_table::scope’ requested
        = interp.require_current_scope ("unpackfields");
          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~
(I'm surprised though that this time through all the other of-* packages built, they almost all failed miserably before on one of the machines)
gdb failed becaue of :
...
make[4]: *** No rule to make target '../bfd/libbfd.a', needed by 'gdb'.
make[4]: *** No rule to make target '../opcodes/libopcodes.a', needed by 'gdb'.
make[4]: Target 'all' not remade because of errors.

Any suggestions? The other machine doesn't fail on these (gdb and nsis, I haven't built package binaries on the other one, and only accidentally let it slip on this one), so I must have setup dependencies inconsistently? I'll move the successful octave-4.3.0+.tar.gz to the other machine which successfully does make nsis-installer and see if I can combine them. But I'd still like to solve this so I can reliably build all the way through on both machines individually.

Do you have the latest mxe-octave? JWE pushed a patch for of-general a few days ago that may have fixed the issues for that package.

For gdb, there was probally a previous error in the build that prevented bfd/libbfd.a etc from being built that may be actual cause of the error

reply via email to

[Prev in Thread] Current Thread [Next in Thread]