[Pkg-octave-devel] octave-forge packages uploaded

Thomas Weber thomas.weber.mail at gmail.com
Sun Dec 14 21:59:56 UTC 2008


On Sun, Dec 14, 2008 at 09:36:12PM +0100, Rafael Laboissiere wrote:
> * Thomas Weber <thomas.weber.mail at gmail.com> [2008-12-14 17:42]:
> 
> > I've uploaded the last missing packages, with a few exceptions:
> > 	optiminterp
> > 	octgpr
> > 	statistics
> 
> Thanks for accomplishing this herculean task.
> 
> > optiminterp is the easist case, we already have the latest version :)
> > octgpr's demos trigger a bug in Octave 3.0.3. We could disable the demo,
> > but Jaroslav pushed a fix into the stable 3.0 repository and plans to
> > release 3.0.4 before Christmas. As I'm likely to forget re-enabling the
> > demo, I'd say we wait for 3.0.4. 
> > 
> > The statistics package is pretty broken in the released 1.0.7 version
> > (filename don't match with functions, ...). There has been quite some
> > work in SVN, though, so I'm planning to upload an SVN snapshot after
> > 3.0.4.
> 
> Yes, let us wait until 3.0.4 comes out.
> 
> An aside question: is it yet the case that when Octave-Forge is released for
> a new Octave version, every package gets a version number bump, even though
> no code is changed for some of them?

Yes. The glue code used in Octave-Forge is a mixture of about
everything you can imagine: Makefiles, Shell scripts, Perl, M4, ...

I don't think it's possible to get this 'fixed' in any way with the old
code. A clean rewrite in just one language seems a much better approach
to me.[1]
Ultimately, I don't think Sourceforge's infrastructure provides enough
room for the quite different things octave-forge needs. 

[1] This is not meant as an insult to the different guys who coded this.
Given the restrictions on Sourceforge and the changing of Octave-Forge
over the years, it's probably the best one could achieve.

	Thomas



More information about the Pkg-octave-devel mailing list