Re: [Pkg-octave-devel] [RFU] octave3.1_3.1.54-2
On Wed, Mar 18, 2009 at 2:45 AM, John W. Eaton <jwe@jweaton.org> wrote:
> On 18-Mar-2009, Rafael Laboissiere wrote:
>
> | * John W. Eaton <jwe@jweaton.org> [2009-03-17 20:59]:
> |
> | > I hope to make a new snapshot soon (3.1.55).
> |
> | Okay. Do you have any idea of when it will happen?
>
> I'll do my best to make it happen within the next 7-10 days.
>
> | > We found some unpleasant bugs that no one seems interested in fixing,
> | > so there probably won't be a 3.0.4 release now.
> |
> | Which are these bugs?
>
> I'm not sure precisely. Maybe Jaroslav can comment.
>
Those were reported in this thread:
http://www.nabble.com/octave-3.0.4-rc5-td22177145.html#a22201560
I see it has grown; so maybe there's a patch now. I'll try going
through. In any case, this again demonstrates 2 things:
1. If we are to keep the current system of releases, there should be
more people able to transplant patches to the stable branch. Thomas
Weber already said it's no problem (the repos are just symlinked to
home dirs). I can still do the physical release.
2. We need more frequent stable forks. In fact this could be
determined by the amount of merging needed when transplanting patches.
In September, they mostly went smoothly. In December, many were
failing. My feeling was that the splitting point happened in early
November; I even pled for a new fork but didn't seem to have convinced
anyone, because most others were worried about the quality.
An alternative is to come up with some totally different scheme.
A "feature freeze" state also doesn't seem to work well for us; people
keep coming with good ideas, bug reports or patches, and putting them
off means risking that they pile up or get forgotten or so.
I think the best scheme is just take one of the snapshot that seems to
be in a good shape and fork 3.2 version from it. Maybe the next one?
Just when it seems the stable branch is lagging behind, we prepare a
new fork. I don't think there's really need to wait for certain
features to be complete - not when the cycle is short enough.
just my 2 cents
regards
--
RNDr. Jaroslav Hajek
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz
Reply to: