osc to stop commit to pkg if there are pending submissions

classic Classic list List threaded Threaded
9 messages Options
Reply | Threaded
Open this post in threaded view
|

osc to stop commit to pkg if there are pending submissions

Tomas Chvatal
Hi all,

as $SUBJECT says it all we in pack team would like to have commit abort
in case there are pending submission against the package. [1]

This means that by accident we would not break pending submissions we
missed and should reduce annoyance level among contributors where they
would have to rebase their work.

There would of course be --force option to disregard this stopper.

Anyone against such change?

Cheers

Tom

[1] https://github.com/openSUSE/osc/pull/337

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Olaf Hering-2
On Wed, Oct 11, Tomas Chvatal wrote:

> Anyone against such change?

Why not fix the actual bug, instead of papering over it?

AFAIK a submit request refers to a very specific state of the sources,
not "tip". Take these two SRs as example:

https://build.opensuse.org/request/show/533379
https://build.opensuse.org/request/show/533381

Here rev#2 and rev#3 are submitted. Not #1, not #N.
Why would that stop anyone from doing further work on a pkg and create
rev#4 at any time?

And since Factory has Staging projects there is even less reason to
reject a submission just because further work was done "at the wrong
time".

Olaf

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Odp.: [opensuse-factory] osc to stop commit to pkg if there are pending submissions

Tomas Chvatal
Olaf Hering píše v St 11. 10. 2017 v 14:53 +0200:

> On Wed, Oct 11, Tomas Chvatal wrote:
>
> > Anyone against such change?
>
> Why not fix the actual bug, instead of papering over it?
>
> AFAIK a submit request refers to a very specific state of the
> sources,
> not "tip". Take these two SRs as example:
>
> https://build.opensuse.org/request/show/533379
> https://build.opensuse.org/request/show/533381
>
> Here rev#2 and rev#3 are submitted. Not #1, not #N.
> Why would that stop anyone from doing further work on a pkg and
> create
> rev#4 at any time?
>
> And since Factory has Staging projects there is even less reason to
> reject a submission just because further work was done "at the wrong
> time".
>
Mostly you will get source conflict in spec or changelog. Regardless
how to implement it you can have scenarios where you won't conflict
indeed but mostly you will. As such I consider it good manners to first
review the incoming requests and then merge it on my side rather than
ask people to redo their work.

Also this is cheap and easy solution that actually prevents the trouble
of disgruntled contributors that have to rework their things, while if
you wanna you can use the -f and keep rolling like Today.

Tom

signature.asc (849 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Dominique Leuenberger / DimStar
In reply to this post by Tomas Chvatal
On Wed, 2017-10-11 at 14:08 +0200, Tomas Chvatal wrote:

> Hi all,
>
> as $SUBJECT says it all we in pack team would like to have commit
> abort
> in case there are pending submission against the package. [1]
>
> This means that by accident we would not break pending submissions we
> missed and should reduce annoyance level among contributors where
> they
> would have to rebase their work.
>
> There would of course be --force option to disregard this stopper.
>
> Anyone against such change?
I'm afraid we will punishg the wrong people - where the one that should
be punished to letting requests linger are the package maintainers, you
now punish any contributor that happened to be slower to submit
anyhthing before somebody else did.

The maintainers will still let the stuff linger around.

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Olaf Hering-2
In reply to this post by Tomas Chvatal
On Wed, Oct 11, Tomas Chvatal wrote:

> Mostly you will get source conflict in spec or changelog. Regardless

You are right, I had a different scenario in mind. Sorry for that.

So you are proposing kind of a lock for a pkg. The lock is at least one
pending SR. The lock will be cleared by processing all incoming SRs.

If thats really the case, no --force is needed.
'osc rq decline/accept N' is needed before a pkg/prj maintainer can
submit further changes. In this case the work of resolving conflicts can
be moved to the pkg maintainer, instead of always forcing it to the
submitter.

Olaf

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Jan Engelhardt-4
In reply to this post by Dominique Leuenberger / DimStar

On Wednesday 2017-10-11 15:06, Dominique Leuenberger / DimStar wrote:

>On Wed, 2017-10-11 at 14:08 +0200, Tomas Chvatal wrote:
>> Hi all,
>>
>> would like to have commit abort in case there are pending submission
>> against the package. [1]
>>
>I'm afraid we will punishg the wrong people - where the one that should
>be punished to letting requests linger are the package maintainers, you
>now punish any contributor that happened to be slower to submit
>anyhthing before somebody else did.

Are you sure we are all on the same page? I read Tomas's proposal as "punish
maintainers that try to do direct-commits without checking for SRs first"…
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Jan Engelhardt-4
In reply to this post by Tomas Chvatal
On Wednesday 2017-10-11 14:08, Tomas Chvatal wrote:

>as $SUBJECT says it all we in pack team would like to have commit abort
>in case there are pending submission against the package. [1]
>
>This means that by accident we would not break pending submissions we
>missed and should reduce annoyance level among contributors where they
>would have to rebase their work.
>
>Anyone against such change?

What happens if there are multiple competing SRs? Accepting one does
count as a commit, and if committing is inhibited that seems like a
snail biting its own tail.


In other news, was there not a change in OBS to make the patcher a
little more intelligent so that it could merge a set of pure additions
to .changes even when the stock /usr/bin/patch approach would otherwise
fail?
--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]

Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Dominique Leuenberger / DimStar
In reply to this post by Jan Engelhardt-4
On Wed, 2017-10-11 at 16:55 +0200, Jan Engelhardt wrote:

> On Wednesday 2017-10-11 15:06, Dominique Leuenberger / DimStar wrote:
> > On Wed, 2017-10-11 at 14:08 +0200, Tomas Chvatal wrote:
> > > Hi all,
> > >
> > > would like to have commit abort in case there are pending
> > > submission
> > > against the package. [1]
> > >
> >
> > I'm afraid we will punishg the wrong people - where the one that
> > should
> > be punished to letting requests linger are the package maintainers,
> > you
> > now punish any contributor that happened to be slower to submit
> > anyhthing before somebody else did.
>
> Are you sure we are all on the same page? I read Tomas's proposal as
> "punish
> maintainers that try to do direct-commits without checking for SRs
> first"…
Ah - that might be true; indeed...

That would be indeed not the worst of ideas; apologies for not reading
precise enough

Cheers
Dominique

signature.asc (201 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: osc to stop commit to pkg if there are pending submissions

Tomas Chvatal
In reply to this post by Jan Engelhardt-4
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

Jan Engelhardt píše v St 11. 10. 2017 v 16:58 +0200:

> On Wednesday 2017-10-11 14:08, Tomas Chvatal wrote:
>
> >
> > as $SUBJECT says it all we in pack team would like to have commit
> > abort
> > in case there are pending submission against the package. [1]
> >
> > This means that by accident we would not break pending submissions
> > we
> > missed and should reduce annoyance level among contributors where
> > they 
> > would have to rebase their work.
> >
> > Anyone against such change?
>
> What happens if there are multiple competing SRs? Accepting one does 
> count as a commit, and if committing is inhibited that seems like a 
> snail biting its own tail.
>
The sr uses bit different logic so that should still keep working.
Technically if you accept one of the multiple sr#s you will have to
decline all the other "broken" ones prior you being able to commit (or
use force).

But I think it is nice way to make people notice the submissions, esp.
since quite often the emails about submissions are lagging behind. :)

Tom
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
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=X4dp
-----END PGP SIGNATURE-----

--
To unsubscribe, e-mail: [hidden email]
To contact the owner, e-mail: [hidden email]