[RFC] Auto-assign regressions (maintainers, read!)

[RFC] Auto-assign regressions (maintainers, read!)

[Giovanni Bajo]

Hello,

I'm sending this message because we were discussing the issue deep into
another thread on the mailing list.

I (and others) believe that it would be useful to automatically assign a PR
about a regression to the author of the patch which caused/exposed the
regression. Of course, this does not want to be a way to coerce people to
work on something: the maintainer can just unassign the PR if he does not
want/have time to work on that, or if he can't fix the exposed bug for some
reason.

We think this would be valuable for the bugmasters, the RM and the
maintainters themselves. Does anybody who actually produce patches have an
objection against this new policy? Otherwise, I'll prepare a patch to our
bugmastering policies to make it go live in a few days.

Thanks
-- 
Giovanni Bajo

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Richard Guenther]

On Thu, 25 Nov 2004 01:33:02 +0100, Giovanni Bajo
<giovannibajo@libero.it> wrote:
> Hello,
> 
> I'm sending this message because we were discussing the issue deep into
> another thread on the mailing list.
> 
> I (and others) believe that it would be useful to automatically assign a PR
> about a regression to the author of the patch which caused/exposed the
> regression. Of course, this does not want to be a way to coerce people to
> work on something: the maintainer can just unassign the PR if he does not
> want/have time to work on that, or if he can't fix the exposed bug for some
> reason.
> 
> We think this would be valuable for the bugmasters, the RM and the
> maintainters themselves. Does anybody who actually produce patches have an
> objection against this new policy? Otherwise, I'll prepare a patch to our
> bugmastering policies to make it go live in a few days.

Can't we have a caused-by-patch field in bugzilla where we could link the patch
causing the failure and a caused-by field with the author (and
automatically CC'ing
him)?  I don't think it is a great idea overloading the assigned-to field.

Richard.

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Daniel Berlin]

On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:

> Can't we have a caused-by-patch field in bugzilla where we could link the patch
> causing the failure and a caused-by field with the author (and
> automatically CC'ing
> him)?  I don't think it is a great idea overloading the assigned-to field.

Everytime you ask me to add another custom field, God kills a kitten.
Please, think of the kittens.


> 
> Richard.

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Richard Guenther]

On Thu, 25 Nov 2004 11:28:39 -0500, Daniel Berlin <dberlin@dberlin.org> wrote:
> On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:
> 
> > Can't we have a caused-by-patch field in bugzilla where we could link the patch
> > causing the failure and a caused-by field with the author (and
> > automatically CC'ing
> > him)?  I don't think it is a great idea overloading the assigned-to field.
> 
> Everytime you ask me to add another custom field, God kills a kitten.
> Please, think of the kittens.

Don't kittens have nine lifes? :)

Richard.

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Andrew MacLeod]

On Thu, 2004-11-25 at 11:32, Richard Guenther wrote:
> On Thu, 25 Nov 2004 11:28:39 -0500, Daniel Berlin <dberlin@dberlin.org> wrote:
> > On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:
> > 
> > > Can't we have a caused-by-patch field in bugzilla where we could link the patch
> > > causing the failure and a caused-by field with the author (and
> > > automatically CC'ing
> > > him)?  I don't think it is a great idea overloading the assigned-to field.
> > 
> > Everytime you ask me to add another custom field, God kills a kitten.
> > Please, think of the kittens.
> 
> Don't kittens have nine lifes? :)
> 

I believe they have to survive kittenhood and become cats before they
get those extra lives.  Maybe its a graduation present :-)

Andrew

RE: [RFC] Auto-assign regressions (maintainers, read!)

[Dave Korn]

> -----Original Message-----
> From: gcc-owner On Behalf Of Richard Guenther
> Sent: 25 November 2004 16:32
> To: Daniel Berlin
> Cc: Giovanni Bajo; gcc
> Subject: Re: [RFC] Auto-assign regressions (maintainers, read!)
> 
> On Thu, 25 Nov 2004 11:28:39 -0500, Daniel Berlin 
> <dberlin@dberlin.org> wrote:
> > On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:
> > 
> > > Can't we have a caused-by-patch field in bugzilla where 
> we could link the patch
> > > causing the failure and a caused-by field with the author (and
> > > automatically CC'ing
> > > him)?  I don't think it is a great idea overloading the 
> assigned-to field.
> > 
> > Everytime you ask me to add another custom field, God kills 
> a kitten.
> > Please, think of the kittens.
> 
> Don't kittens have nine lifes? :)
> 
> Richard.


  If I were to ask Daniel to add a field for the number of remaining
kitten-lives, would it end up auto-decrementing?


    cheers, 
      DaveK
-- 
Can't think of a witty .sigline today....
 

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Kai Henningsen]

dberlin@dberlin.org (Daniel Berlin)  wrote on 25.11.04 in <1101400119.6587.2.camel@linux.site>:

> On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:
>
> > Can't we have a caused-by-patch field in bugzilla where we could link the
> > patch causing the failure and a caused-by field with the author (and
> > automatically CC'ing
> > him)?  I don't think it is a great idea overloading the assigned-to field.
>
> Everytime you ask me to add another custom field, God kills a kitten.
> Please, think of the kittens.

Death to the kitten-murderer! Become an atheist!

MfG Kai

Re: [RFC] Auto-assign regressions (maintainers, read!)

[bobby mcnulty junior]

[-- Attachment #1: Type: text/plain, Size: 765 bytes --]

Kai Henningsen wrote:
> dberlin@dberlin.org (Daniel Berlin)  wrote on 25.11.04 in <1101400119.6587.2.camel@linux.site>:
> 
> 
>>On Thu, 2004-11-25 at 10:44 +0100, Richard Guenther wrote:
>>
>>
>>>Can't we have a caused-by-patch field in bugzilla where we could link the
>>>patch causing the failure and a caused-by field with the author (and
>>>automatically CC'ing
>>>him)?  I don't think it is a great idea overloading the assigned-to field.
>>
>>Everytime you ask me to add another custom field, God kills a kitten.
>>Please, think of the kittens.
> 
> 
> Death to the kitten-murderer! Become an atheist!
> 
> MfG Kai
> 
God created the kittens in the 1st place.


-- 
Robert McNulty Junior
email: snyder100@bellsouth.net
webpage: www.geocities.com/bobbymcn2004

[-- Attachment #2: bobbymcn.vcf --]
[-- Type: text/x-vcard, Size: 260 bytes --]

begin:vcard
fn:Robert McNulty Junior
n:McNulty Junior;Robert
adr:;;3971 Linda Ann Drive;Millbrook;AL;36054;USA
email;internet:snyder100@bellsouth.net
tel;home:1-334-285-5490
x-mozilla-html:FALSE
url:http://www.geocities.com/bobbymcn2004
version:2.1
end:vcard

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Ranjit Mathew]

Richard Guenther wrote:
> 
> Can't we have a caused-by-patch field in bugzilla where we could link the patch
> causing the failure and a caused-by field with the author (and
> automatically CC'ing
> him)?  I don't think it is a great idea overloading the assigned-to field.

IMHO it'd be better to just have a new keyword instead and put
the link to the offending patch in the body of the PR; just as
it is done with the "patch" keyword for PRs that have a patch
pending review.

Ranjit.

-- 
Ranjit Mathew          Email: rmathew AT gmail DOT com

Bangalore, INDIA.      Web: http://ranjitmathew.tripod.com/

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Diego Novillo]

On Thu, 2004-11-25 at 01:33 +0100, Giovanni Bajo wrote:

> I (and others) believe that it would be useful to automatically assign a PR
> about a regression to the author of the patch which caused/exposed the
> regression.
> 
I don't think it's a good idea.  Assigned-to has different semantics.
It means "working on" or "meaning to work on".  How about CCing the
likely author or adding a new field?


Diego.

Re: [RFC] Auto-assign regressions (maintainers, read!)

[Ian Lance Taylor]

"Giovanni Bajo" <giovannibajo@libero.it> writes:

> I (and others) believe that it would be useful to automatically assign a PR
> about a regression to the author of the patch which caused/exposed the
> regression. Of course, this does not want to be a way to coerce people to
> work on something: the maintainer can just unassign the PR if he does not
> want/have time to work on that, or if he can't fix the exposed bug for some
> reason.

I don't think this would be a good change.

I think there should be a way to identify, in the bugzilla database,
the person who caused a regression, when that is known.  However, I
don't think using the Assigned field is the right mechanism.  As
others have said, it means that we can no longer tell whether somebody
is actually working on the bug.  In a volunteer project, in which some
patches come from people who just e-mail them in, to get approved and
applied by more active maintainers, we can not reasonably expect every
contributor to use bugzilla.

(Daniel, I can supply kittens if you need them.)

Ian