[Pkg-octave-devel] Releasing octave 4.0.0~rc2 to experimental

[Mike Miller]

On Wed, Apr 22, 2015 at 12:35:13 +0200, Sébastien Villemot wrote:
> Le lundi 20 avril 2015 à 13:32 +0200, Rafael Laboissiere a écrit :
> 
> > I investigated a little bit further this issue.  Even though both 3.8.2 
> > and 4.0.0 have the same API number (v49+), .oct files compiled with 3.8.2 
> > do not produce the same results as those compiled with 4.0.0.  For 
> > instance, with the octave-signal package installed (which contains 
> > remez.oct) the following runs perfectly with 3.8.2:
> > 
> >      octave:1> remez (15, [0,0.3,0.4,1], [1,1,0,0]);
> > 
> > However, with 4.0.0, I get:
> > 
> >      octave:1> remez (15, [0,0.3,0.4,1], [1,1,0,0]);
> >      error: remez: number of taps must be an integer greater than 3
> > 
> > I am afraid that we will have o declare Breaks relationships for all 
> > packages containing .oct files.  What do you think?
> 
> I have seen that you have added these Breaks relationships, thanks for
> that.

I remember seeing this kind of behavior when testing packages and
switching between 3.8 and 4.0.

> I am however a bit surprised by the fact that the SOVERSION of liboctave
> has not been bumped upstream. It looks like it should have been, given
> the breakage that you describe.

Agree, this probably should be done. Several deprecated functions and
the deprecated Octave_map class have been dropped from the library in
version 4.0.

-- 
mike