Re: 2.5.5 release soon

On Wed, Apr 06, 2005 at 09:05:42AM +0300, Peter Damoc wrote:
> On Wed, 06 Apr 2005 05:00:58 +0300, Ron <ron@xxxxxxxxxx> wrote:
> 
> >On Tue, Apr 05, 2005 at 02:53:57PM -0700, Robin Dunn wrote:
> >>>I've said it before, I'll say it again... with the exception of few
> >>>preview releases *ALL* wxpython releases from the 2.5 branch proved
> >>>just  as stable as 2.4.2.4 yet many people still remain with 2.4.2.4
> >>>because  that is "the latest stable release".
> >>
> >>Yep.
> >
> >Part of this is a distribution issue.  It may seem initially paradoxical
> >to realise that wxPython at any given moment is generally more stable
> >than the library it is 'built on', but anecdotally it seems to often
> >be the case.
> 
> I think is due to the glue.... a lot of problems get caught in the glue  
> layer ;)

Robin has a new nickname ?  ; )

> >That makes things hard, because pressure comes from the Python community
> >to release the underlying lib before _it_ is ready.
> 
> you'll never be able to make the lib "perfect"...

lately I'll settle for 'compiles'...

> that's why I think  
> scheduled releases make sense. They impose discipline too... (1-2 months  
> API freeze before release would be great).

Don't get me wrong, I love deadlines.  They solve the halting problem.
And like the late D.A. I especially love the whooshing sound they make
as they go by.

But they _don't_ improve quality, or the decisions about how to do that.
And if you'll flip through the last couple of months worth of wx-dev,
you'll see what happens when discipline and pragmatism meet a deadline
in the cold light of day.

> >wx2.5 is certainly not a stable library, however 'bug free' wxPython may
> >be.  You seem to be confusing these two things.  All wxPython releases
> >are **by definition** 'stable' -- ie. we don't change them any more.
> 
> Again with the glue.... I think wxpython "stability" is due to the lack of  
> the low level bugs, is due to the marriage with python ;)

You are still missing the point.  Its "stability" is because Robin will
cluebat anyone who changes the api unnecessarily.  The presence or lack
of bugs at whatever level is irrelevant to that.  libwx is not stable,
its api changes every day, even during a 'freeze', and without consultation
or research before a deadline.

That Robin shields you from this moving target seems to be what you
attribute to 'glue'.

When libwx stops doing this we can call it stable, and only then can
we realistically start to count the number of remaining bugs and
assess its suitability for production use.  Its got nothing to do with
being perfect, but there is a little more finesse involved than just
drawing a line in the sand and calling it release day.

There are a lot more factors at play than you so far give credit to.

> >Hopefully the C++ library team will move into the mode of testing and
> >careful changes that have been employed here for the last several months
> >once it is declared 'stable' too.  At least the last minute API changes
> >should slow a little I hope, and the chances of a tagged release actually
> >building will increase...
> 
> sometimes too much freedom can hurt too. Dictatorship is good if the  
> dictator is a benevolent one ;)

I'll take a good and fair 'king' over a bureau of thugs any day, but you
can't have it both ways.  You either need to dictate or listen.  I
haven't seen too many swords pulled from rocks around here lately :-)

> I don't understand how project management issues can "degrade into a  
> flaming tornado of death" when is crystal clear,

Subscribe to wx-dev and give it a whirl ...   But its on your own neck
if you do.  I won't feed that fire.

> at least in my view, the  
> fact that Robin has the best overview of the wx project since he works  
> full time on it, his opinion should weight a lot more that the opinion of  
> any other developer.

Even if it is dumb?   That is just stupid.  His opinion should weigh on
its own merit.  There are plenty of reasons that Robin's opinion is well
received by many, including myself, without needing to stoop to such
shallow means of picking an arbiter.

> I'm not saying that he should slice and dice as he  
> pleases BUT his position should prevent the flaming. Let them vote on it  
> if they think themselves as democratic.

Like the U.S. we never profess to be democratic ourselves.  We're a
bunch of rabid egos all chasing different bunnies.  We'll listen to
you if you are smart or one of the few that actually do write code
that is useful to others, but we never promised to give you a vote,
just the freedom to take the source and do it your way, any time you
think ours isn't working anymore.

But that takes real leadership, and real vision.  The trouble with
dictators these days is they all seem to want to come into the job
straight out of school.  Who's going to unite behind a stand-up
dictator with nothing of substance to dictate?

If you want to change this, there will need to be a coup.  Whether or
not it is bloodless will show whether or not you were qualified to
lead it... :-)

If voting changed anything, it would be illegal.

  Ron