RM responsibility

[Kent Gibson]

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



Aaron Bentley wrote:
> Kent Gibson wrote:
>> I'm tempted to suggest that cores should also require a review by
>> 2 others, i.e their don't get to review their own code, but that
>> would just add to the review backlog.
>
> What motivates that suggestion?
Could be demonic possession, or just my evil streak showing.

But my thinking is it would level the playing field for cores and
non-cores.
Finding a core to do a review can be difficult, especially when big
changes are under way, such as the performance drive now and the
dirstate stuff a couple of releases back.
A core only needs to get one other core to do a review.  That is much
easier than finding two.  Additionally, non-cores have no leverage to
get a core's attention, while cores have at least some.
All in all it feels much harder for someone outside the core to get a
patch approved.
And that is a really bad thing cos it undermines community involvement.

Btw, the same motivation is behind my original suggestion that the RM
push the review process along, effectively playing proxy for the
non-cores.  I prefer that solution to this one since it's lower overhead.

>
> I guess my feeling is that you get diminishing returns from reviews
> the more experience the submitter has.  Certainly you get fewer
> style issues, and it tends to come down to "is this a good idea"
> rather than "is this well-implemented" more often.
>
I agree totally on the style issues.
But if it is just a matter of getting the idea okayed, then it sounds
like a good reason for a review (of the idea) by more than just one
other core, and before coding even starts.  I guess that is starting
to head down the blueprint path, though more lightweight.

>> It would be nice for the cores to aim to have code reviewed
>> within
> a week.
>> That would certainly help to keep the pipeline flowing and keep
>> the non-cores more engaged.
>
> I do try to do that.  Some patches lie outside my comfort zone, so
> they either take longer, or I leave them to someone else.  And
> there are some patches I don't want to review.  It shouldn't fall
> to me all the time, but it often seems to.
>

I know you do.  You already do more than your fair share of reviewing,
IMO.  e.g. I think you have reviewed ALL my patches, and John has
reviewed more than his fair share as well.  I'm impressed by the work
you guys do.
I'm not trying to apportion blame, just suggesting that striving for
such a goal would help improve overall efficiency and community
involvement.

Cheers,
Kent.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGi87LgoxTFTi1P8QRAlNsAKCd6jpKOcqMPwW7Dga/86E6J2kM+ACg72Ms
ND5zqGcrFl+uZntIfFB0yY4=
=W9gp
-----END PGP SIGNATURE-----