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

[Antoine Pitrou]

Hi Victor, hi Raymond,

Le 20/01/2017 à 11:45, Victor Stinner a écrit :
> 
> Inaccurate summary: I'm a strong supporter of "it's better to ask
> forgiveness than permission", whereas Raymond considers that I
> introduced too many regressions with my workflow.

It's a matter of balance.  "Easier to ask forgiveness than permission"
is fine for small-scale changes with little ramification or impact on
other parts.  It certainly wouldn't be appropriate for PEP-sized changes.

It's true that there has been a lot of churn lately, which is a bit
difficult to follow.  OTOH, I don't follow much anyway, because I'm
unmotivated and don't care enough except for very few areas.

> Since one or maybe even two years, I noticed that many of my issues
> were blocked by the lack of reviews. As you wrote, only few developer
> have the knowledge and background to be able to provide a good review
> (not only "tests pass, so LGTM") on my changes modifying the Python
> core.

Yes, this is still an important issue.  There are topics on which there
is almost never anybody available to do a review, so sometimes you may
have to move forward and cross fingers that you're not breaking anything.

> It's strange, but the process of opening an issue and attaching the
> patch usually helps to review the code myself (find bugs, or more
> generally enhance the patch).

Ha, that's exactly the same thing here.  Reading my patch in a Web
browser makes me likely to spot problems that I wouldn't have found
otherwise.

> By the way, it's painful to squash a long patch serie into a giant
> patch, much harder to review, where changes don't make sense at all at
> the first look. Again, a better reviewing tool supporting patch series
> (GitHub) will help here too.

I've never done patch series.  They look very painful to maintain,
because tooling for them is very primitive (whether with hg or git, AFAIK).

> Oh by the way, when I read your comment, I understand that I'm
> responsible of all regressions. It's true that I introduced
> regressions, that's where I said "shit happens" (or more politically
> correct: "it's better to ask forgiveness than permission" ;-)).

If what was lacking was more reviewing of patches properly posted to the
tracker, then you're not the only person responsible, we're all
collectively responsible.  Chastising one of the few persons who puts in
a lot of work is unfair.

Regards

Antoine.