[gentoo-dev] keywording of new style virtuals

[gentoo-dev] keywording of new style virtuals

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 362 bytes --]

All,

from my previous discussion, I am about to put a new virtual in the
tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
would normally use even though the default package the virtual will
bring in is stable everywhere?

I'm thinking I can take the virtual straight to stable in this situation,,
but I want to be sure.

William


[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] keywording of new style virtuals

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 736 bytes --]

On 06.07.2011 21:55, William Hubbs wrote:
> All,
> 
> from my previous discussion, I am about to put a new virtual in the
> tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
> would normally use even though the default package the virtual will
> bring in is stable everywhere?
> 
> I'm thinking I can take the virtual straight to stable in this situation,,
> but I want to be sure.
> 

Nothing stable should be using the virtual at the time of commit so
what's the benefit of going stable fast? Repoman should also be
preventing commits straight to stable. I would stable the virtual at the
same time as someone stable starts to use it (which probably means the
30 day period).

Regards,
Petteri


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-dev] keywording of new style virtuals

[William Hubbs]

[-- Attachment #1: Type: text/plain, Size: 1253 bytes --]

On Wed, Jul 06, 2011 at 10:17:28PM +0300, Petteri Räty wrote:
> On 06.07.2011 21:55, William Hubbs wrote:
> > All,
> > 
> > from my previous discussion, I am about to put a new virtual in the
> > tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
> > would normally use even though the default package the virtual will
> > bring in is stable everywhere?
> > 
> > I'm thinking I can take the virtual straight to stable in this situation,,
> > but I want to be sure.
> > 
> 
> Nothing stable should be using the virtual at the time of commit so
> what's the benefit of going stable fast? Repoman should also be
> preventing commits straight to stable. I would stable the virtual at the
> same time as someone stable starts to use it (which probably means the
> 30 day period).

Actually we could use it faster than that. I want to add a
virtual/service-manager (see http://bugs.gentoo.org/373843) for
sys-apps/openrc and sys-apps/systemd then add it to the system set, so
it would be used immediately everywhere.

That will also make it possible for folks using systemd to remove
openrc from their systems if they want to do so, which they
can't right now because baselayout has a PDEPEND on openrc.

William

[-- Attachment #2: Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] keywording of new style virtuals

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 1470 bytes --]

On 06.07.2011 22:45, William Hubbs wrote:
> On Wed, Jul 06, 2011 at 10:17:28PM +0300, Petteri Räty wrote:
>> On 06.07.2011 21:55, William Hubbs wrote:
>>> All,
>>>
>>> from my previous discussion, I am about to put a new virtual in the
>>> tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
>>> would normally use even though the default package the virtual will
>>> bring in is stable everywhere?
>>>
>>> I'm thinking I can take the virtual straight to stable in this situation,,
>>> but I want to be sure.
>>>
>>
>> Nothing stable should be using the virtual at the time of commit so
>> what's the benefit of going stable fast? Repoman should also be
>> preventing commits straight to stable. I would stable the virtual at the
>> same time as someone stable starts to use it (which probably means the
>> 30 day period).
> 
> Actually we could use it faster than that. I want to add a
> virtual/service-manager (see http://bugs.gentoo.org/373843) for
> sys-apps/openrc and sys-apps/systemd then add it to the system set, so
> it would be used immediately everywhere.
> 
> That will also make it possible for folks using systemd to remove
> openrc from their systems if they want to do so, which they
> can't right now because baselayout has a PDEPEND on openrc.
> 

I don't see why one would need to hurry with changing the system set. On
the contrary I would proceed more cautiously than usual.

Regards,
Petteri


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

Re: [gentoo-dev] keywording of new style virtuals

[Samuli Suominen]

On 07/06/2011 09:55 PM, William Hubbs wrote:
> All,
> 
> from my previous discussion, I am about to put a new virtual in the
> tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
> would normally use even though the default package the virtual will
> bring in is stable everywhere?
> 
> I'm thinking I can take the virtual straight to stable in this situation,,
> but I want to be sure.
> 
> William
> 

You can grab both stable and ~arch KEYWORDS="" from the deps listed in
|| ( ) for the virtual and commit it with repoman --force.

Then the virtual is instantly available for usage.

- Samuli

Re: [gentoo-dev] keywording of new style virtuals

[Ulrich Mueller]

>>>>> On Wed, 6 Jul 2011, William Hubbs wrote:

> from my previous discussion, I am about to put a new virtual in the
> tree. Do I need to use the same ~arch/30 day wait/stabilize cycle I
> would normally use even though the default package the virtual will
> bring in is stable everywhere?

> I'm thinking I can take the virtual straight to stable in this
> situation,, but I want to be sure.

I don't think that we have fixed rules about this, but there was a
discussion in 2007:
<http://archives.gentoo.org/gentoo-dev/msg_682b30a8c8aa0b29602645271b19f603.xml>

- Virtuals don't install any files, so there's no need for a 30 days
  waiting period. They can go straight to stable if at least one
  dependency is stable (for the arch in question).
- Noone from arch teams objected about virtuals being stabilised by
  their maintainer.

Ulrich