Re: MacOS X 10.3.9 build but load-save.m test failure

Ok, I believe I've found the problem, with Dmitri's help... The issue is 
that the type of the saved variable is saved as an 8-bit value, I was 
reading it into a 4-byte variable and then using X_CAST to get to the 
save_type. The 3 used bytes then came into play... This would also cause 
problems on little endian systems where the size of the atrix or the 
number of non-zero terms was 256 or greater....

The attached patch should fix this problem....

D.

--

-- 
David Bateman                                David.Bateman <at> motorola.com
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

Index: src/ov-cx-sparse.cc
===================================================================
RCS file: /usr/local/cvsroot/octave/src/ov-cx-sparse.cc,v
retrieving revision 1.5
diff -c -r1.5 ov-cx-sparse.cc

Re: MacOS X build but load-save.m test failure

On 29/04/2005, at 3:45 AM, John W. Eaton wrote:

> | > | FAIL: octave.test/io/load-save.m
> | >
> | > Can you please investigate what is causing this test to fail?
> |
> | Log entry:
> | Running ../../test/octave.test/io/io.exp ...
> | EXEC: ../src/octave -f -q -p .:../src//:../../test/../scripts//
> | ../../test/octave.test/io/load-save.m
> |
> | Octave Output:
> | error: load: trouble reading binary file `binary.mat'
> | error: evaluating for command near line 77, column 5
> | error: evaluating if command near line 67, column 3
> | error: called from `testls'
> | error: near line 183 of file `../../test/octave.test/io/load-save.m'
> | child process exited abnormally
> | FAIL: octave.test/io/load-save.m
>
> I can load the binary.mat file that you included in the attachment
> with a copy of Octave built from the current CVS sources.  So if there
> is a bug, I'm not sure what it is.  It would help if someone who can
> reproduce the failure could do some debugging.
>

I have now rebuilt Octave using gfortran from gcc-4.0.0 release on  
MacOS X 10.4 Tiger and I have exactly the same problem.

Do you have any suggestions about how to troubleshoot it?

Re: MacOS X build but load-save.m test failure

On 02/05/2005, at 6:08 PM, Bill Northcott wrote:
>
> I have now rebuilt Octave using gfortran from gcc-4.0.0 release on  
> MacOS X 10.4 Tiger and I have exactly the same problem.
>

Just to set the record straight.  It won't actually build with the  
gcc-4.0.0 release code you need to apply the patch for PR13082.   
Other wise you will die with an alternate entry point problem.  The  
patch is now committed so it should build with gfortran from the next  
gcc release.

Bill Northcott

-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------

Re: MacOS X 10.3.9 build but load-save.m test failure

On 28/04/2005, at 2:05 PM, Dmitri A. Sergatskov wrote:
>> the patch.  At least we should see what happens in the 4.0.x bug-fix
>> releases, and maybe even wait until 4.1.  If anyone has any influence
>> over gfortran development, it would be helpful to try to get some
>> attention given to the bug report for ENTRY statements in functions.
>>
>
> I am sorry -- I meant the message to Bill Northcott. May be it is
> easier to apply the patch (temporary) and build with pure Apple
> toolchain than try to circumvent problems with different compilers.
>
> As far as I can tell from:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13082
>
> the bug has been fixed in the release version.
> I am waiting for RH to put out the RPMs...

For the record this is NOT fixed in the gcc-4.00 code release.  It  
has been fixed in the cvs as of 29 April 2005.  So it should be fixed  
in the next release.

I have manually applied the patch to Apple's gcc-5013 and that seems  
to work for functions with alternate entry points.

Bill Northcott

-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org

Build problem on MacOS X 10.4 Tiger

There is a problem with the configuration script for Octave on MacOS  
X 10.4 Tiger.

The new release of MacOS includes the BSD libedit library linked as  
libreadline.  The configure script picks this up instead of the GNU  
readline installed in /usr/local and dies.

I fixed the problem by shamelessly borrowing the readline autoconf  
macro from R.  This macro properly distinguishes between BSD and GNU  
libreadline even when both are present.

My new definition of OCTAVE_ENABLE_READLINE in aclocal.m4 is as follows.

AC_DEFUN(OCTAVE_ENABLE_READLINE, [
   USE_READLINE=true
   LIBREADLINE=
   AC_ARG_ENABLE(readline,
     [  --enable-readline       use readline library (default is yes)],
     [if test "$enableval" = no; then
        USE_READLINE=false
        warn_readline="command editing and history features require  
GNU Readline"
      fi])
      if $USE_READLINE; then
       AC_CHECK_HEADERS(readline/history.h readline/readline.h)
       r_save_LIBS="${LIBS}"
       LIBS=
       ## don't use the cached value as we need to rebuild LIBS
       unset ac_cv_lib_readline_rl_callback_read_char
       AC_CHECK_LIB(readline, rl_callback_read_char)

Re: MacOS X build but load-save.m test failure

Bill Northcott wrote:

> On 29/04/2005, at 3:45 AM, John W. Eaton wrote:
>
>> | > | FAIL: octave.test/io/load-save.m
>> | >
>> | > Can you please investigate what is causing this test to fail?
>> |
>> | Log entry:
>> | Running ../../test/octave.test/io/io.exp ...
>> | EXEC: ../src/octave -f -q -p .:../src//:../../test/../scripts//
>> | ../../test/octave.test/io/load-save.m
>> |
>> | Octave Output:
>> | error: load: trouble reading binary file `binary.mat'
>> | error: evaluating for command near line 77, column 5
>> | error: evaluating if command near line 67, column 3
>> | error: called from `testls'
>> | error: near line 183 of file `../../test/octave.test/io/load-save.m'
>> | child process exited abnormally
>> | FAIL: octave.test/io/load-save.m
>>
>> I can load the binary.mat file that you included in the attachment
>> with a copy of Octave built from the current CVS sources.  So if there
>> is a bug, I'm not sure what it is.  It would help if someone who can
>> reproduce the failure could do some debugging.
>>
>
> I have now rebuilt Octave using gfortran from gcc-4.0.0 release on  
> MacOS X 10.4 Tiger and I have exactly the same problem.

Re: MacOS X build but load-save.m test failure

David Bateman wrote:
...
> 
> Can you try the CVS of octave. I commit some patches, but have no 
> feedback on if it fully addresses the load-save issues...

It works. Thank you, David.

GNU Octave, version 2.9.2 (sparc-sun-solaris2.8).
Copyright (C) 2005 John W. Eaton.
....

octave:1> load test11.mat
octave:2> whos

*** local user variables:

   Prot Name            Size                     Bytes  Class
   ==== ====            ====                     =====  =====
    rw- __nargin__      1x1                          8  scalar
    rwd a10             2x7                         14  string
    rwd a11             2x2                          4  int8 matrix
    rwd a12             2x2                          8  int16 matrix
    rwd a13             2x2                         16  int32 matrix
    rwd a14             2x2                         32  int64 matrix
    rwd a15             2x2                          4  uint8 matrix
    rwd a16             2x2                          8  uint16 matrix
    rwd a17             2x2                         16  uint32 matrix
    rwd a18             2x2                         32  uint64 matrix
    rwd a19           100x100                     1604  sparse matrix

error using plot3

Bug report for Octave 2.1.57 configured for i686-pc-linux-gnu

Description:
-----------

When I try to use plot3 the following error occurs.

parse error near line 93 of file 
/usr/share/octave/2.1.57/site/m/octave-forge/plot/plot3.m

 >>>     endwhile
               ^

error: `for' command matched by `endwhile' near line 93 column 5
error: `plot3' undefined near line 3 column 4

Repeat-By:
---------
z = [0:0.05:5];
plot3(cos(2*pi*z), sin(2*pi*z), z, ";helix;");

This is the example given by "help plot3".

Configuration (please do not edit this section):
-----------------------------------------------

uname output:     Linux nibblerx 2.6.9-gentoo-r9 #19 Wed Apr 27 12:53:35 
CEST 2005 i686 Intel(R) Pentium(R) 4 Mobile CPU 1.70GHz GenuineIntel 
GNU/Linux
configure opts:   '--disable-static' '--enable-shared' '--enable-dl' 

Re: MacOS X build but load-save.m test failure

On 02/05/2005, at 9:13 PM, David Bateman wrote:
>> I have now rebuilt Octave using gfortran from gcc-4.0.0 release  
>> on  MacOS X 10.4 Tiger and I have exactly the same problem.
>>
>> Do you have any suggestions about how to troubleshoot it?
>>
>> Bill Northcott
>>
>>
>
> Can you try the CVS of octave. I commit some patches, but have no  
> feedback on if it fully addresses the load-save issues...
>

Attempt to build cvs code died thus:
powerpc-apple-darwin8-g++-4.0.0: spec failure: unrecognized spec  
option 'Q'
g++ -c  -I. -I../../src -I.. -I../liboctave -I../src -I../libcruft/ 
misc -I../.. -I../../liboctave -I../../src -I../../libcruft/misc  - 
DHAVE_CONFIG_H  -Wall -W -Wshadow -g -O2 ../../src/pager.cc -o pager.o
powerpc-apple-darwin8-g++-4.0.0: spec failure: unrecognized spec  
option 'Q'
g++ -c  -I. -I../../src -I.. -I../liboctave -I../src -I../libcruft/ 
misc -I../.. -I../../liboctave -I../../src -I../../libcruft/misc  - 
DHAVE_CONFIG_H  -Wall -W -Wshadow -g -O2 parse.cc -o parse.o
powerpc-apple-darwin8-g++-4.0.0: spec failure: unrecognized spec  
option 'Q'
y.tab.c: In function 'int octave_parse()':
y.tab.c:3443: error: expected primary-expression before '__attribute__'
y.tab.c:3443: error: expected `;' before '__attribute__'

Unidentified subject!

To: bug <at> octave.org
Cc: etienne
Subject: pause() doesn't pause (2.1.69)
--------
Bug report for Octave 2.1.69 configured for i386-pc-linux-gnu

Description:
-----------

  pause() doesn't wait for a key to be pressed.

Repeat-By:
---------

  title ("entropy");plot (1:10,1:10); pause;title ("population"); plot (1:100,(1:100).^2); 

Fix:
---

  * If possible, replace this item with a description of how to
    fix the problem (if you don't have a fix for the problem, don't
    include this section, but please do submit your report anyway).

Configuration (please do not edit this section):
-----------------------------------------------

uname output:     Linux madrid 2.4.22-1-686-smp #5 SMP Sat Oct 4 14:35:05 EST 2003 i686 GNU/Linux
configure opts:   '--prefix=/usr' '--datadir=/usr/share' '--libdir=/usr/lib'
'--libexecdir=/usr/lib' '--infodir=/usr/share/info' '--mandir=/usr/share/man'
'--with-blas=-lblas-3' '--with-lapack=-llapack-3' '--with-hdf5' '--with-fftw'