[pypy-dev] pypy-sync thursday 5:30 GMT+2

[holger krekel]

Hi Armin, 

On Thu, Apr 20, 2006 at 13:24 +0200, Armin Rigo wrote:
> On Tue, Apr 18, 2006 at 10:37:31AM +0200, holger krekel wrote:
> > - what needs to be done until Iceland (21st May) for 0.9? 
> 
> A possible missing clasification on the issue tracker is the separation
> between the two goals of that release: one is community-oriented, and
> the other is to fullfill EU requirements.  For the latter, we are mostly
> done apart from tasklet pickling.

I disagree.  We cannot just ship the current svn/pypy/dist and 
i actually think it's very misleading to give this impression. 
Getting to 0.9 is more than just a few days work or even (as you somewhat 
imply apart from tasklet pickling) no work at all.  And don't forget
the reports for that matter. 

> This is the one that needs to find a
> leader *now*.  It's a contractual thing.  I agree that all other points
> are very nice and some are rather needed, but we are still allowed to
> miss one of them, like weakrefs, __subclasses__, or moving gc's...

sure, but it's not that we don't have tons of other stuff
waiting for us after 0.9.  Generally i think we should be 
careful if we further postpone issues into the last 7 months of
the EU project (where people might get the funny idea to even take
holidays) that got already postponed for 5-9 months. 
but i am fine with missing _some_ of the issues. 

> My point is that I would personally like some more focus on new
> experimental features instead of spending too much time on polishing,
> making performance more stable, even more compliance, etc. -- I think
> that's what we really promized to both the community and the EU.  I'm
> not trying to minimize the importance of all that work, which will have
> to be done sooner or later, but basically we promized many great things
> so the sooner we show at least experimentally that they are possible,
> the more interested people will become IMHO.

Hum, but especially users of the release ("the community" resp.)
will want to get some reasonably stable and entry points 
for using PyPy (especially the ext-compiler) which requires some
polishing and documentation work, doesn't it?   

> I hope this doesn't sound too negative!  It was not my intention.  My
> intention is just to reiterate what I see as the mid-term guidelines
> that I need to keep in mind.
> 
> > (...)
> >   - adding missing support (if any) for app level stackless module
> >     (e.g. yield_current_frame_to_caller ?) 
> 
> No, this was about the higher-level interfaces -- we cannot expose
> yield_current_frame_to_caller directly.  It's about exposing channels to
> make tasklets usable, and greenlets.

sure, i didn't mean to expose yield_current_frame_to_caller at app-level 
but to support it for the stackless (interp-level) module which itself 
supports tasklets etc. 

    holger