octave-maintainers
[Top][All Lists]
Advanced

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

Re: 3.0.4 RC7


From: Rafael Laboissiere
Subject: Re: 3.0.4 RC7
Date: Sat, 28 Mar 2009 00:14:55 +0100
User-agent: Mutt/1.5.18 (2008-05-17)

* Jaroslav Hajek <address@hidden> [2009-03-27 13:14]:

> the 3.0.4 RC7 tarballs are available at the usual location:
> http://artax.karlin.mff.cuni.cz/~hajej2am/ulozna/octave/
> 
> build reports welcome

Build and runs fine on Debian sid.  Thanks for that, as usual your
efforts are highly appreciated.

There is still a bug in this version, though.  I mentioned it in a previous
thread [1] and it has been also reported by a Debian user [2].  This bug is
exposed by:

octave3.0:1> n=100; E=eig(randn(n,n));
error: memory exhausted or requested size too large for range of Octave's index 
type -- trying to return to prompt

It a weird bug with unknown origin.  It only happens for me on amd64, not
on i386.  On amd64, it only shows up when I am in a sid chroot

Anyway, this bug was also present in 3.1.54 but disappeared in 3.1.55.
John Eaton thinks that a patch against src/EIG.cc [3] fixes the problem.
I adapted this patch for 3.0.4 and attached it below.  It is a one-line
patch and I think it is worth applying it to the final 3.0.4 tarball.

[1] 
http://www-old.cae.wisc.edu/pipermail/octave-maintainers/2009-March/011308.html
[2] http://bugs.debian.org/521345
[3] http://hg.savannah.gnu.org/hgweb/octave/rev/728e7943752d

-- 
Rafael

Attachment: idx-not-volatile-in-eig.diff
Description: Text Data


reply via email to

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