[Sugar-devel] [SWEETS] glucose-0.92 via sweets

[Aleksey Lim]

On Sat, Jul 16, 2011 at 06:56:30PM +0000, Aleksey Lim wrote:
> On Sat, Jul 16, 2011 at 01:36:32PM +0100, Daniel Drake wrote:
> > On 16 July 2011 12:50, Aleksey Lim <alsroot at activitycentral.org> wrote:
> > > All of them on bugs.sl.o, since I'm not maintaing collab code and it was rewritten in 0.9x
> > > (and never worked fine since then for me), plus jabber.sl.o has bunch of noise trafic,
> > > I, w/ tweaking prosody patch, patched client code as well. ie, final
> > > implementaion might be too different. thats why these patches are on
> > > bugs.sl.o.
> > 
> > I'd suggest that you submit them as patches to the mailing list -
> > thats the current community norm, even if you aren't a maintainer, and
> > will form the discussions that would help turn them into a final
> > implementation.
> 
> You got me wrong, dunno for others but I'm prefering working in two
> modes, patching stable code (and emailing to sugar-devel@) and devepping
> new features (and do not email every commit to ML).
> 
> In my mind new collab code was never stable but unfortunately
> committed to the trunk. In that case I was very surprised when tried recent
> 0.92, that OLPC is going release these days, and found bunch of critical issues
> (and got assumption that if olpc is ok w/ that, they patched ejabberd,
> which is abs. wrong way for me).
> 
> Personally, if I would glucose maintainer (but I'm not), I will fallback to PS
> and form new collab code as a feature w/ people who will work in
> development mode (ie, not sending every commit to ML) and make it stable
> (being assured by tests). Then, it might replace PS.

Besides, the one of accept criterias needs to be hight level client API/behaviour
description. It should simplify support on both, client and server
sides. The exact issue w/ new code is that it behaves differently to PS.

> Thats my personal vision on new collab feature. To move forward, and have it
> in 0.92 if people want, we need not maintainers (and every commit
> reviewers/accepters on ML) but collab developers who will work in
> development mode to revisit and understand(since we don't have original
> feature developer) the entirely implementation.
> 
> > 
> > Thanks for your work.
> > Daniel
> > 
> 
> -- 
> Aleksey

-- 
Aleksey