[gentoo-dev] Documenting touching of arch profiles' files

[gentoo-dev] Documenting touching of arch profiles' files

[Michael Palimaka]

Hi all,

With regards to bug #304435[1], we would like to formalise the policy 
for touching arch profiles' files.

The key suggested points:

* Archs profiles should generally only be touched by members of that 
arch team, unless prior permission is given

* Exception: anyone may add a mask to an arch profile only if
	- it delays visibility of something new for that arch (eg. dependencies 
introduced in a version bump), and
	- it is not reasonable to follow the standard keyword dropping 
procedure (many other packages would be affected), and
	- the responsible arch team is not responsive

* The person touching arch profiles is responsible for the subsequent 
maintenance of said entries, and any subsequent breakage.

Thoughts?

Best regards,
Michael

[1]: https://bugs.gentoo.org/show_bug.cgi?id=304435

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Robin H. Johnson]

On Fri, Nov 02, 2012 at 12:47:34AM +1100, Michael Palimaka wrote:
> Hi all,
> 
> With regards to bug #304435[1], we would like to formalise the policy 
> for touching arch profiles' files.
> 
> The key suggested points:
Ok, this then clears the way for MySQL 5.5 to enter the tree.

bug #351931: dev-util/systemtap, needs ~arm ~hppa ~ia64 ~sparc
bug #429710: dev-util/google-perftools, needs ~alpha ~arm ~hppa ~ia64 ~ppc ~sparc
bug #429708: dev-libs/jemalloc, needs ~alpha ~ia64 ~sparc

I'm going to put the following masks in for the above:
alpha:
>=dev-db/mysql-5.5 tcmalloc jemalloc
>=dev-db/mariadb-5.5 tcmalloc jemalloc

arm:
>=dev-db/mysql-5.5 systemtap tcmalloc
>=dev-db/mariadb-5.5 systemtap tcmalloc

hppa:
>=dev-db/mysql-5.5 systemtap tcmalloc
>=dev-db/mariadb-5.5 systemtap tcmalloc

ia64:
>=dev-db/mysql-5.5 systemtap tcmalloc jemalloc
>=dev-db/mariadb-5.5 systemtap tcmalloc jemalloc

ppc/ppc64:
>=dev-db/mysql-5.5 tcmalloc
>=dev-db/mariadb-5.5 tcmalloc

sparc:
>=dev-db/mysql-5.5 systemtap tcmalloc jemalloc
>=dev-db/mariadb-5.5 systemtap tcmalloc jemalloc

-- 
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Diego Elio Pettenò]

On 01/11/2012 16:47, Robin H. Johnson wrote:
> I'm going to put the following masks in for the above:

If you're already doing the job would you mind just masking the use flag
globally, and not just for mysql/mariadb?

-- 
Diego Elio Pettenò — Flameeyes
flameeyes@flameeyes.eu — http://blog.flameeyes.eu/

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Robin H. Johnson]

On Thu, Nov 01, 2012 at 04:51:52PM -0700, Diego Elio Pettenò wrote:
> On 01/11/2012 16:47, Robin H. Johnson wrote:
> > I'm going to put the following masks in for the above:
> If you're already doing the job would you mind just masking the use flag
> globally, and not just for mysql/mariadb?
I did package.use.mask already, too late.

-- 
Robin Hugh Johnson
Gentoo Linux: Developer, Trustee & Infrastructure Lead
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11ACBA4F 4778E3F6 E4EDF38E B27B944E 34884E85

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Markos Chandras]

On Thu, Nov 1, 2012 at 1:47 PM, Michael Palimaka <kensington@gentoo.org> wrote:
> Hi all,
>
> With regards to bug #304435[1], we would like to formalise the policy for
> touching arch profiles' files.
>
> The key suggested points:
>
> * Archs profiles should generally only be touched by members of that arch
> team, unless prior permission is given
>
> * Exception: anyone may add a mask to an arch profile only if
>         - it delays visibility of something new for that arch (eg.
> dependencies introduced in a version bump), and
>         - it is not reasonable to follow the standard keyword dropping
> procedure (many other packages would be affected), and
>         - the responsible arch team is not responsive
>
> * The person touching arch profiles is responsible for the subsequent
> maintenance of said entries, and any subsequent breakage.
>
> Thoughts?
>
> Best regards,
> Michael
>
> [1]: https://bugs.gentoo.org/show_bug.cgi?id=304435
>

As Tommy[D] pointed out in IRC, developers are free(wrong word?) to
touch package.use.mask for their packages but they should get an ACK
for use.mask or just tell arches to do it on their behalf. This is an
addition to what you have already said above.

-- 
Regards,
Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Michał Górny]

[-- Attachment #1: Type: text/plain, Size: 1788 bytes --]

On Fri, 2 Nov 2012 09:50:24 +0000
Markos Chandras <hwoarang@gentoo.org> wrote:

> On Thu, Nov 1, 2012 at 1:47 PM, Michael Palimaka <kensington@gentoo.org> wrote:
> > Hi all,
> >
> > With regards to bug #304435[1], we would like to formalise the policy for
> > touching arch profiles' files.
> >
> > The key suggested points:
> >
> > * Archs profiles should generally only be touched by members of that arch
> > team, unless prior permission is given
> >
> > * Exception: anyone may add a mask to an arch profile only if
> >         - it delays visibility of something new for that arch (eg.
> > dependencies introduced in a version bump), and
> >         - it is not reasonable to follow the standard keyword dropping
> > procedure (many other packages would be affected), and
> >         - the responsible arch team is not responsive
> >
> > * The person touching arch profiles is responsible for the subsequent
> > maintenance of said entries, and any subsequent breakage.
> >
> > Thoughts?
> >
> > Best regards,
> > Michael
> >
> > [1]: https://bugs.gentoo.org/show_bug.cgi?id=304435
> >
> 
> As Tommy[D] pointed out in IRC, developers are free(wrong word?) to
> touch package.use.mask for their packages but they should get an ACK
> for use.mask or just tell arches to do it on their behalf. This is an
> addition to what you have already said above.

What about eclass-wide USE_EXPAND flags? I have recently added masks to
the PYTHON_TARGETS for Python implementation not being keyworded on
particular arches.

With the exception of hppa which explicitly says its use.mask shouldn't
be touched without permission, and now I can't enable pypy on flaggie
because that arch is slacking. Great, isn't it?

-- 
Best regards,
Michał Górny

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 316 bytes --]

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Markos Chandras]

On Fri, Nov 2, 2012 at 9:58 AM, Michał Górny <mgorny@gentoo.org> wrote:
> On Fri, 2 Nov 2012 09:50:24 +0000
> Markos Chandras <hwoarang@gentoo.org> wrote:
>
>> On Thu, Nov 1, 2012 at 1:47 PM, Michael Palimaka <kensington@gentoo.org> wrote:
>> > Hi all,
>> >
>> > With regards to bug #304435[1], we would like to formalise the policy for
>> > touching arch profiles' files.
>> >
>> > The key suggested points:
>> >
>> > * Archs profiles should generally only be touched by members of that arch
>> > team, unless prior permission is given
>> >
>> > * Exception: anyone may add a mask to an arch profile only if
>> >         - it delays visibility of something new for that arch (eg.
>> > dependencies introduced in a version bump), and
>> >         - it is not reasonable to follow the standard keyword dropping
>> > procedure (many other packages would be affected), and
>> >         - the responsible arch team is not responsive
>> >
>> > * The person touching arch profiles is responsible for the subsequent
>> > maintenance of said entries, and any subsequent breakage.
>> >
>> > Thoughts?
>> >
>> > Best regards,
>> > Michael
>> >
>> > [1]: https://bugs.gentoo.org/show_bug.cgi?id=304435
>> >
>>
>> As Tommy[D] pointed out in IRC, developers are free(wrong word?) to
>> touch package.use.mask for their packages but they should get an ACK
>> for use.mask or just tell arches to do it on their behalf. This is an
>> addition to what you have already said above.
>
> What about eclass-wide USE_EXPAND flags? I have recently added masks to
> the PYTHON_TARGETS for Python implementation not being keyworded on
> particular arches.
>
> With the exception of hppa which explicitly says its use.mask shouldn't
> be touched without permission, and now I can't enable pypy on flaggie
> because that arch is slacking. Great, isn't it?
>
> --
> Best regards,
> Michał Górny

As Michael already said on the very first post on this thread, you are
free to touch the file is arch is slacking.

-- 
Regards,
Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Ben de Groot]

On 2 November 2012 18:01, Markos Chandras <hwoarang@gentoo.org> wrote:
> On Fri, Nov 2, 2012 at 9:58 AM, Michał Górny <mgorny@gentoo.org> wrote:
[...]
>> With the exception of hppa which explicitly says its use.mask shouldn't
>> be touched without permission, and now I can't enable pypy on flaggie
>> because that arch is slacking. Great, isn't it?
>
> As Michael already said on the very first post on this thread, you are
> free to touch the file is arch is slacking.

In that case this whole policy is unnecessary, as the minor arches are
always slacking and unresponsive, while x86 and amd64 have no problems
with developers doing what they need to do in their profiles.

In my opinion we should simply state that:
1) contacting arch teams is preferred, but should not hold up
development activity if they are not immediately responsive
2) we need make sure a bug is filed for each issue, so arch teams are
kept in the loop, and everyone can track what is going on
3) small changes are no problem (e.g. package.use.mask), but
wider-reaching changes should be announced / discussed in advance

-- 
Cheers,

Ben | yngwin
Gentoo developer
Gentoo Qt project lead, Gentoo Wiki admin

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Jeroen Roovers]

On Thu, 1 Nov 2012 23:47:58 +0000
"Robin H. Johnson" <robbat2@gentoo.org> wrote:

> On Fri, Nov 02, 2012 at 12:47:34AM +1100, Michael Palimaka wrote:
> > Hi all,
> > 
> > With regards to bug #304435[1], we would like to formalise the
> > policy for touching arch profiles' files.
> > 
> > The key suggested points:
> Ok, this then clears the way for MySQL 5.5 to enter the tree.
> 
> bug #351931: dev-util/systemtap, needs ~arm ~hppa ~ia64 ~sparc
> bug #429710: dev-util/google-perftools, needs ~alpha ~arm ~hppa ~ia64
> ~ppc ~sparc bug #429708: dev-libs/jemalloc, needs ~alpha ~ia64 ~sparc
> 
> I'm going to put the following masks in for the above:

Sending mail to a general mailing list is not the same as notifying an
arch team.


     jer

Re: [gentoo-dev] Documenting touching of arch profiles' files

[Markos Chandras]

On Fri, Nov 2, 2012 at 3:25 PM, Ben de Groot <yngwin@gentoo.org> wrote:
> On 2 November 2012 18:01, Markos Chandras <hwoarang@gentoo.org> wrote:
>> On Fri, Nov 2, 2012 at 9:58 AM, Michał Górny <mgorny@gentoo.org> wrote:
> [...]
>>> With the exception of hppa which explicitly says its use.mask shouldn't
>>> be touched without permission, and now I can't enable pypy on flaggie
>>> because that arch is slacking. Great, isn't it?
>>
>> As Michael already said on the very first post on this thread, you are
>> free to touch the file is arch is slacking.
>
> In that case this whole policy is unnecessary, as the minor arches are
> always slacking and unresponsive, while x86 and amd64 have no problems
> with developers doing what they need to do in their profiles.
>
> In my opinion we should simply state that:
> 1) contacting arch teams is preferred, but should not hold up
> development activity if they are not immediately responsive
> 2) we need make sure a bug is filed for each issue, so arch teams are
> kept in the loop, and everyone can track what is going on
> 3) small changes are no problem (e.g. package.use.mask), but
> wider-reaching changes should be announced / discussed in advance
>

which are the minor arches? And how do you classify arches on major and minor?

I already said that if an arch is not responsive go ahead and commit
your changes.

Discussing it in a public ML everytime you want to touch the use.mask
of an arch is not efficient. A bug may be preferred yes.
Like I said, we already touch package.use.mask without previously
getting an ACK from the arch so the policy for this
situation will not change.

-- 
Regards,
Markos Chandras / Gentoo Linux Developer / Key ID: B4AFF2C2