[python-committers] My cavalier and aggressive manner, API change and bugs introduced for basically zero benefit

[INADA Naoki]

I read this whole thread.  Here is my opinion.


1. Lack of reviewers

It's really problem.  If well written patch is not reviewed and
committed for long time,
patch writer (core developer, or potential core developer) may lost
their motivation.
Motivated contributor will be lower and lower.

I wrote compact dict patch in June, before Python 3.6a3.
Luckily, the patch was reviewed and committed in September, right before 3.6b1.
But I realized Python lacks active reviewers heavily.
That's main reason why I wanted to be core developer.

I was very lucky.  But there might be some contributors lost
motivation because of
lack of review.


2. Stability

Stability is important.  Review is valuable for stability.
But even though careful review, regression may be happens.

Python 3.6 has many improvements.  And many of them are committed
right before 3.6b1, at sprint.
http://blog.python.org/2016/09/python-core-development-sprint-2016-36.html

I feel adding fundamental changes (like FASTCALL) at early stage of
development process (alpha1~3) is also valuable for stability of last
beta, RC and final.

And I think advocating running test with "nightly" Python on Travis-CI
may help stability,
even though there are few beta users.

One regression caused by compact dict (*), and it was one reason of
3.6rc2 was released.
The regression is found by py.test.

*) An excuse: the issue was not from compact dict.  It was issue of
key sharing dict.
   But compact dict increased method which can touch the issue, and
py.test touched it.


3. Domain expert

I agree that consult to expert is good.
But sometimes the expert may not be active.

And mechanical changes by sed or Coccinelle may affects wide part.
It's not realistic to consult all experts.

We should think carefully it's worth enough.  But after some devs agree
it's worth enough, "ask forgiveness than permission" seems realistic approach,
like Stefan reverted _decimal change.


4. Github

Github may ease some problems we have, especially reviewing large patch.
https://github.com/blog/2123-more-code-review-tools

Python 3.4.6 and 3.5.3 are released already, migration happens near future?
Anyway, I'm thanks to people working on the migration.


Regards,