Re: Classdef for next release

[Rik]

On 07/30/2013 08:40 AM, address@hidden wrote:
> Message: 6
> Date: Tue, 30 Jul 2013 11:40:18 -0400
> From: Michael Goffioul <address@hidden>
> To: "John W. Eaton" <address@hidden>
> Cc: octave maintainers mailing list <address@hidden>
> Subject: Re: merge classdef to default?
> Message-ID:
>       <address@hidden>
> Content-Type: text/plain; charset="iso-8859-1"
>
> On Tue, Jul 30, 2013 at 11:19 AM, John W. Eaton <address@hidden> wrote:
>
>> > On 07/30/2013 09:57 AM, Michael Goffioul wrote:
>> >
>> >  There's preliminary support in the classdef branch. It started here:
>>> >>
>>> >> http://hg.savannah.gnu.org/**hgweb/octave/rev/7368654f302f?**revcount=240<http://hg.savannah.gnu.org/hgweb/octave/rev/7368654f302f?revcount=240>
>>> >>
>> >
>> > Should we merge the classdef branch to default?
>> >
>> > If we do it now, does it cause any harm to have it in the upcoming
>> > release?  I suppose that since it is incomplete, people will try it and
>> > have some trouble, then report bugs.  But we can (try to) make it clear
>> > that it is a preliminary work in progress.
>> >
>> > If you think it will cause more harm than good, then we can delay the
>> > merge until after we merge default to stable for the release.
> That's a difficult decision. But if the classdef support was not part of
> the original feature list, I'd say leave it for later. I don't think  it
> would cause any harm to merge, most of the code is independent from the
> rest of octave, except for the package support which touched the load-path
> and symbol-table aspects. The problem is indeed to have only partial
> classdef support and advertise it. I'm not opposed to the merge, bug if
> it's to get tons of bug report about things not being
> supported/implemented, then no thanks.
>
> Maybe we could add a big fat warning, whenever some aspects of classdef are
> being used, that say this is WIP and incomplete, please don't report any
> bugs about unimplemented features?

I'd also recommend waiting.  I think we already have more than enough
changes to make 3.8 an interesting release.  Bumping the major version to
4.0 also seems like a good point to introduce things like classdef which
are rather more revolutionary than the incremental changes within the
current 3.X branch.

--Rik