3.2.0 release candidate 7

[Jaroslav Hajek-2]

hi all,

Octave 3.2.0 RC7 tarballs available at:
http://artax.karlin.mff.cuni.cz/~hajej2am/ulozna/octave/

changes since RC6 are only cosmetic (plus one Mac OS X specific
issue). If you have sucessfully tested RC5 or RC6, there's little need
to do again, unless you're interested in trying out the cosmetic changes.
Unless anyone reports significant problems, RC7 will become 3.2.0 on Friday.

PS. the school server artax only offers limited capacity to students,
so I only keep the most recent pair of RC tarballs. If anyone wishes
to keep the whole RC series, please suggest an alternative storage
location.


changeset:   9291:0dc1d0f1f79c
user:        Jaroslav Hajek <highegg@...>
date:        Wed Jun 03 13:51:41 2009 +0200
summary:     omit conf.texi from distribution

changeset:   9292:aa2e52bd82dd
user:        Soren Hauberg <hauberg@...>
date:        Wed May 27 14:10:32 2009 +0200
summary:     Fix call to page_screen_output in the Bugs chapter of the manual

changeset:   9293:d371cb65428a
user:        Soren Hauberg <hauberg@...>
date:        Wed May 27 14:35:11 2009 +0200
summary:     Fix output from 'print_usage' in Errors chapter in the manual

changeset:   9294:f7a520eb6325
tag:         qparent
tag:         3-2-0-rc7
user:        Rob Mahurin <rob@...>
date:        Wed Jun 03 08:03:59 2009 -0400
summary:     src/Makefile.in: Add CARBON_LIBS to OCTINTERP_LINK_DEPS.


--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

[Benjamin Lindner]

Jaroslav Hajek wrote:
> hi all,
>
> Octave 3.2.0 RC7 tarballs available at:
> http://artax.karlin.mff.cuni.cz/~hajej2am/ulozna/octave/
>

Hi,

The changeset in
http://www.nabble.com/msvc-specific-linker-flags-in-mkoctfile.cc.in-to23670863.html
has not been commented on yet.

reasonable chance to get it in to 3.2 ?

benjamin

[Jean-Francois Cardoso]

Thanks to John and Jaroslav for explaining the BLAS problem reported earlier.

I build 3.2.0 RC 7 on MacOS, but I see a lot of test failures,
all related to structs, and apparently to struct arrays.

I append below the relevant parts of fntests.log
and the result of octave_config_info.

Let me know if more info could be useful.

Cheers, JF

fntests.logcfginfo

[Jaroslav Hajek-2]

On Thu, Jun 4, 2009 at 5:46 PM, Jean-Francois Cardoso<cardoso@...> wrote:

>
> Thanks to John and Jaroslav for explaining the BLAS problem reported earlier.
>
> I build 3.2.0 RC 7 on MacOS, but I see a lot of test failures,
> all related to structs, and apparently to struct arrays.
>
> I append below the relevant parts of fntests.log
> and the result of octave_config_info.
>
> Let me know if more info could be useful.
>
> Cheers, JF
>
> http://www.nabble.com/file/p23872602/fntests.log fntests.log
> http://www.nabble.com/file/p23872602/cfginfo cfginfo

Interesting. Since nobody else sees those failures, I guess you'll
need to do some investigation what's wrong. Some of the errors
resemble problems fixed long ago - are you sure you didn't mix up some
sources?
If not, my very wild guess is that maybe your g++ somehow manages to
miscompile Octave. 4.2.1 is somewhat old; maybe there were some bugs?

Try isolating a single piece of code that does not work, for instance
a(1).x.x = 1; a(2).x.x = 2;
a

put it in a script, then try running with ./run-octave -valgrind and
check for invalid memory accesses and the like.
if it does not help, you can try to debug Octave (./run-octave -g).

regards

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

[Jaroslav Hajek-2]

On Wed, Jun 3, 2009 at 8:39 PM, Benjamin Lindner<lindnerben@...> wrote:

> Jaroslav Hajek wrote:
>>
>> hi all,
>>
>> Octave 3.2.0 RC7 tarballs available at:
>> http://artax.karlin.mff.cuni.cz/~hajej2am/ulozna/octave/
>>
>
> Hi,
>
> The changeset in
> http://www.nabble.com/msvc-specific-linker-flags-in-mkoctfile.cc.in-to23670863.html
> has not been commented on yet.
>
> reasonable chance to get it in to 3.2 ?
>

I applied it to Savannah; will be in 3.2.1. It would be preferable,
however, if such patches were handled by people actively working with
Windows.

regards

--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz

[Jean-Francois Cardoso]

Jaroslav Hajek-2 wrote:

On Thu, Jun 4, 2009 at 5:46 PM, Jean-Francois Cardoso<[e-mail]> wrote:
>
> Thanks to John and Jaroslav for explaining the BLAS problem reported earlier.
>
> I build 3.2.0 RC 7 on MacOS, but I see a lot of test failures,
> all related to structs, and apparently to struct arrays.
>
> I append below the relevant parts of fntests.log
> and the result of octave_config_info.
>
> Let me know if more info could be useful.
>
> Cheers, JF
>
> http://www.nabble.com/file/p23872602/fntests.log fntests.log
> http://www.nabble.com/file/p23872602/cfginfo cfginfo

Interesting. Since nobody else sees those failures, I guess you'll
need to do some investigation what's wrong. Some of the errors
resemble problems fixed long ago - are you sure you didn't mix up some
sources?
If not, my very wild guess is that maybe your g++ somehow manages to
miscompile Octave. 4.2.1 is somewhat old; maybe there were some bugs?

Try isolating a single piece of code that does not work, for instance
a(1).x.x = 1; a(2).x.x = 2;
a

put it in a script, then try running with ./run-octave -valgrind and
check for invalid memory accesses and the like.
if it does not help, you can try to debug Octave (./run-octave -g).

Hi Jaroslav,
To make sure I was not messing with the sources,
I have re-built everything starting from the tar balls.
The result is that the same build script (appended below)
produces a working octave 3.0.5 but a faulty octave 3.2.0.

I am not skilled enough to use dbg or valgrind
but I may try upgrading gcc and gfortran from 4.2 to 4.4
and see if it solves the struct array problem (without
destroying my system ;-)

I will report on this later today if the upgrade goes well.

Cheers, JF



#!/bin/bash
# Build script for octave

## Pick one
version=3.1.55 ; release=""
version=3.0.4 ; release="-rc7"
version=3.0.5 ; release=""
version=3.2.0 ; release="-rc7"

tarball=octave-$version$release.tar.gz

rm -fR cd octave-$version/
tar xzvf $tarball
cd octave-$version
export CPPFLAGS="-I/usr/local/include/readline"
export LDFLAGS="-L/usr/local/lib"
export FFLAGS="-ff2c" ; ## for octave 3.2.0

###  Test gcc 4.2.   Where has gcc_select gone ?
export CC=gcc-4.2
export CPP=cpp-4.2
export CXX=g++-4.2
export CXXCPP=cpp-4.2 ##      C++ preprocessor ????????

./configure
make -j2
make check