Commits to master in my absence

Commits to master in my absence

[Richard Purdie]

I'm a little frustrated to see what happened when I was away from a
couple of days. I thought we'd agreed we'd queue up patches in a branch
and then I'd take care of things when I got back.

I appreciate a serious bug was then found and a fix was committed
against what I thought was agreed. This then opened the gates and more
patches were merged with decreasing severity and in the end something
got added which is just plain broken :(.

The problematic patch is:

http://git.openembedded.org/cgit.cgi/openembedded-core/commit/?id=e23f9ce928353c9da2b9c4bfa9a1a125a7d160f2

and I've commented in emails elsewhere against the various attempts to
further "fix" this.

I *really* need people to step up and start thinking about quality and
the long term impact of changes rather than just X fixes bug Y which I'm
hitting right now. This is particularly true if anyone wants to step
into the position of doing final commit merging. You *have* to take the
time to fully understand and review the changes you're adding and
consider the big picture.

I'm really hoping that I can find people who I can share the final
review/commit responsibility with but until people stop presenting me
patches which I have to reject we're not going to get the point where
they can work in that role :(.

This is particularly frustrating since I've seen people push patches
they know to be lacking in quality just because they were running low on
time and hoped they might get away with it.

Cheers,

Richard (a bit disappointed)

Re: Commits to master in my absence

[Khem Raj]

On (01/08/11 14:45), Richard Purdie wrote:
> I'm a little frustrated to see what happened when I was away from a
> couple of days. I thought we'd agreed we'd queue up patches in a branch
> and then I'd take care of things when I got back.
> 
> I appreciate a serious bug was then found and a fix was committed
> against what I thought was agreed. This then opened the gates and more
> patches were merged with decreasing severity and in the end something
> got added which is just plain broken :(.
> 
> The problematic patch is:
> 
> http://git.openembedded.org/cgit.cgi/openembedded-core/commit/?id=e23f9ce928353c9da2b9c4bfa9a1a125a7d160f2
> 
> and I've commented in emails elsewhere against the various attempts to
> further "fix" this.
> 
> I *really* need people to step up and start thinking about quality and
> the long term impact of changes rather than just X fixes bug Y which I'm
> hitting right now. This is particularly true if anyone wants to step
> into the position of doing final commit merging. You *have* to take the
> time to fully understand and review the changes you're adding and
> consider the big picture.
> 
> I'm really hoping that I can find people who I can share the final
> review/commit responsibility with but until people stop presenting me
> patches which I have to reject we're not going to get the point where
> they can work in that role :(.
> 
> This is particularly frustrating since I've seen people push patches
> they know to be lacking in quality just because they were running low on
> time and hoped they might get away with it.
> 
> Cheers,
> 
> Richard (a bit disappointed)

It was me.
since master was broken for arm (even qemuarm) after the tune overhaul
It was a mimimum workaround since many devs were complaining about it I
thought until we rework the tune overhaul this could be a survival path
and at the same time tune overhaul for arm could be worked out. I was
just thinking about making devs keep going.
But I apologize for this. Kindly revert it.

-Khem

Re: Commits to master in my absence

[Tom Rini]

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

On 08/01/2011 06:45 AM, Richard Purdie wrote:

> I'm a little frustrated to see what happened when I was away from a
> couple of days. I thought we'd agreed we'd queue up patches in a branch
> and then I'd take care of things when I got back.

So, I agree with everything you did say.  But I'd also like to suggest
we should have gone with "merge everything Monday" not "merge everything
Wednesday before RP runs off".  Yes, this would have meant either moving
some testing, or having testing also pull from a different (merged)
branch.  And I too was "wait, what, more stuff went in?  I thought we
caught the big problem already".

- -- 
Tom Rini
Mentor Graphics Corporation
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJONv2eAAoJEI4NMjfc3nJxudcIAMg7yMIdl648ZdRsFGAUTKTD
HSP8lLzawtEcl1bxektShH7amCf45wHvlw36joNNt3dG78FwAfO9bozubZEypwcS
N7YXBo8JVJVi1WUIni23CZor4I091Ubt0tEHFOk4iDIyU9B9q+T77xg2qn+r6RJN
7yNzs7XRffM8D2v7OxG/6kK4FvQDflVbnqEYkN+dE0rgUFOMUDdaSkB5JO0NShG2
yYLkXljRNJ24Of5crsuQ/F4Vqk86b41Ax9jYrdFUOl+qgdovbPs3ninKtbDgT9ez
Adh8behhx/LvbQQNldOi0l/UakPAmEPHEjEBNxmKX/GoZjfLHN0nb4Ucy5sAtBM=
=PIo2
-----END PGP SIGNATURE-----

Re: Commits to master in my absence

[Richard Purdie]

On Mon, 2011-08-01 at 12:25 -0700, Tom Rini wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On 08/01/2011 06:45 AM, Richard Purdie wrote:
> 
> > I'm a little frustrated to see what happened when I was away from a
> > couple of days. I thought we'd agreed we'd queue up patches in a branch
> > and then I'd take care of things when I got back.
> 
> So, I agree with everything you did say.  But I'd also like to suggest
> we should have gone with "merge everything Monday" not "merge everything
> Wednesday before RP runs off".  Yes, this would have meant either moving
> some testing, or having testing also pull from a different (merged)
> branch.  And I too was "wait, what, more stuff went in?  I thought we
> caught the big problem already".

If I'd merged everything unexpectedly and not communicated the situation
I think that would be bad. As it was, I did merge things but I also
hopefully clearly communicated the issues and the expectation. Master is
the development tip and some instability, particularly as new features
merge is not entirely unexpected at this point in the cycle.

I'd hope we don't end up in a situation like that again, if we do, I'll
see what other options there and weigh them up as the situation
dictates.

Cheers,

Richard

Re: Commits to master in my absence

[Tom Rini]

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

On 08/01/2011 12:56 PM, Richard Purdie wrote:
> On Mon, 2011-08-01 at 12:25 -0700, Tom Rini wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On 08/01/2011 06:45 AM, Richard Purdie wrote:
>>
>>> I'm a little frustrated to see what happened when I was away from a
>>> couple of days. I thought we'd agreed we'd queue up patches in a branch
>>> and then I'd take care of things when I got back.
>>
>> So, I agree with everything you did say.  But I'd also like to suggest
>> we should have gone with "merge everything Monday" not "merge everything
>> Wednesday before RP runs off".  Yes, this would have meant either moving
>> some testing, or having testing also pull from a different (merged)
>> branch.  And I too was "wait, what, more stuff went in?  I thought we
>> caught the big problem already".
> 
> If I'd merged everything unexpectedly and not communicated the situation
> I think that would be bad. As it was, I did merge things but I also
> hopefully clearly communicated the issues and the expectation. Master is
> the development tip and some instability, particularly as new features
> merge is not entirely unexpected at this point in the cycle.
> 
> I'd hope we don't end up in a situation like that again, if we do, I'll
> see what other options there and weigh them up as the situation
> dictates.

Yes, it was an intentional and communicated decision.  I guess what my
gut says right now is that while some folks are willing to live with
some "it's broken, try again in a few days or go back to ...", a lot of
the community expects things to be mostly working, and there was also
some "ok, I want to expand / use this stuff, but..." that we ran into,
which really is situation dependent.  And yes, next time we will need to
weigh the options once again.

- -- 
Tom Rini
Mentor Graphics Corporation
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQEcBAEBAgAGBQJONwn7AAoJEI4NMjfc3nJx2vQH/2u2MKHdqhsHNfoZoHlhPCJQ
jk144kwp69HHVS9gKjiJCf1fkum9MTblT5V5VDIYgoD2yHDBnCe/P/9GWS8rlTcq
lUweqPMDuWvoWPEGRWnNLPAnXP+bsWqoQLGEUuHkcmMag4yJsJSZYTmU6btTOSNa
ON6Q4RnkTcD08AiHark/R5YHS46HaQ6tVyRpK3DgMUgfGVGnYa0EFzGG73p6q44x
GKsXqNRverf4uu9KbnRTcciuxeIytGlZ3ewT9yHJBcKOjaT9a8FEkA03fHOIqRy/
VJnfjOL4NbdoSRa67INQm8nAnI20WzxCrkzZPJ/2+VXXCfrZ4TFiabUviocPpmA=
=oTtt
-----END PGP SIGNATURE-----