[development] Listen

[Gerhard Killesreiter]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Dries Buytaert schrieb:
> 
> On 02 Jul 2007, at 21:05, Augustin (Beginner) wrote:
>> My main point is: please do listen to people who are better known than
>> I am,
>> when they talk about some overdue systemic changes.
> 
> The main reason that keeps patches from getting committed faster is the
> lack of good reviews.  The main challenge is to increase the amount and
> the quality of patch reviews and to reduce the number of silly "+1"s.
> 
> People posting a "+1" waste a lot of people's time -- it makes dozens of
> people recheck the issue, and it does not buy you any more respect or
> trust.

Correct.

>  If we can stop posting "+1"s (or "subscribe"s for that matter),

We should go ahead and convert issues to comments and then install a
subscribe module on drupal.org. Derek, what's the status on converting
issues to comments? We should mail all the people who do "subscribe"s to
help with that issue...

> that would save me some time, it would increase the signal to noise
> ratio and it would avoid the false sense of support.

Ack.

> We should also change the perception that RTBC means "a core committer
> needs to look at this".

Well, when we created that status, we actually meant it to mean exactly
that. The false understanding is that "nobody else needs to look at this".

> When a patch is RTBC, it still means that
> everyone needs to look at it, and that's part of the reason why many
> patches are still in the RTBC queue.

Right. Including on occassion patches for an unsupported release. :p

>  I'll try to be faster to send back
> these to the "code needs (better) review" status, if that helps.

I believe it does, yes.

> Ultimately, this is something everyone can help with.  If a patch is in
> RTBC for too long, it probably means it could use more quality reviews.

Yeah, but this is difficult to guess.

> Maybe we need a 'decay feature' that sets a patch back to 'code needs
> review' after 2 weeks as RTBC?

That's an interesting idea. However, I've just tested some of the oldest
RTBC patches and they were still good. These were small bugfix patches,
though.

> During the next couple of weeks, I'll pay close attention to my workflow
> and usage patterns surrounding the issue queues.  I'll try to gather
> some statistics of why patches are rejected, and how much time is spent
> doing what.  How frequently do I revisit an issue, and how often that
> means something useful was added to the issue?  Things like that.  I'll
> also keep an eye open for things that would help us.

The outcome of this will be interesting.

> Automatically checking whether a patch still applies would be useful
> already.

Guess we should think about implementing it, then.

Cheers,
	Gerhard

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFGiiTQfg6TFvELooQRAuMrAJ0UtqUFrtWl84p/eUO0R09RdHZWvACfZ1fo
bPEiWh5OrX2xxCgKCYc0kSY=
=vI/m
-----END PGP SIGNATURE-----