python-dev Summary for 2004-08-01 through 2004-08-15

[Robert Brewer]

Brett C. wrote:
> I just don't want people to suddenly get the impression that they can
> always just throw a little mutiny every time Guido makes a decision. 
> Do that and he will just ignore what people want period and never even
> give people a chance to propose an alternative once it reaches
> python-dev.

Well said.

> And I want people to realize my tirade was just that, a personal
> tirade.  It was out of frustration and thus should be taken with a
> grain of salt.

But that frustration is grounded in reality.

> And that frustration came from the feeling that the Python community,
> which I love, seemed to suddenly form a mob, grab pitchforks and
> torches, and started screaming.  And then they seemed to hold a public
> meeting with *everyone* proposing their idea and not completely
> listening to initial decisions.

I attribute the breakdown-of-process to scale issues, and not much else.
The decorator debate had:
    a) the largest body of dissenters I've ever seen for a new Python
feature, and
    b) the largest volume of alternate proposals.

Many new features have had a certain amount of tweaking to be done even
after Guido has pronounced. Each of these tweaks requires communication.
In the case of decorators, the volume of communication required was
simply too much for the medium of Usenet, which promotes short,
tangential discussions rather than systemic analysis. This is the
primary reason I felt a focused paper would benefit the process.

One of my hopes is that, for large, complex community responses in the
future, someone else will do the same. This proposal sets a precedent
for those cases which overload the normal response process. However, it
should be made clear that _not every response warrants this format_.

In Brett's language, go ahead and hold "public meetings". But either
hire a Usenet police force or (much better) police yourselves--when the
debate begins to explode, form a task force and write a proposal or two.
I am *not* placing blame on anyone here, just pointing out that this is
the first time such a step was needed. We are all watching it play out
for the first time. Let's learn from it and apply those lessons next
time.

Although I like J2, I'm not rabid about it (and despite the opinions of
the pundits, I do *not* have "too much free time"; far from it). I could
have written an equally-compelling paper on any of the alternatives. I
am far more dedicated to the "meta-outcome" of this proposal, that Guido
is presented with alternatives in a usable and effective fashion, than I
am in which option he selects. The "next level" of course, is whether
that goal is met and how to reproduce such a success (or avoid such a
failure) in the future.

> I truly hope that next time Guido makes a decision that people as a
> whole disagree with everyone involved can get together and discuss it
> calmly without flooding my inbox.  =)

Me too. This isn't a fundamentally hard problem, it simply happens to be
a management issue for a bunch of techies. Once in a while, Guido's
"cabinet" needs to manage people and process as much as it does source
code. As Python gains more of a following, this will occur more often.


Robert Brewer
MIS
Amor Ministries
fumanchu at amor.org