[gentoo-project] EAPI bump should require revbump

[gentoo-project] EAPI bump should require revbump

[Ian Stakenvicius]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hey all -- there was just a discussion in #gentoo-dev about this, so
following up here..

Due to the fact that an EAPI-bump can imply different behaviour from
eclasses or the PM, an EAPI-bump should in most cases also require an
ebuild revbump.

Why i'm bringing this up here, is because Chansaw and I were wondering
if common sense will not be enough to ensure this and it should be
made a policy to revbump when migrating to a new EAPI ?

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlCBWwsACgkQ2ugaI38ACPDq9QEAv4dtB8q5h4k6P5Fcl5KJ6AZz
+stbD96s7kNG5cLu1zcBAJ/iyTOvAH5GIYhnSDKOrz2JODD7Zq0II9W+7L4XMfQa
=03Pd
-----END PGP SIGNATURE-----

Re: [gentoo-project] EAPI bump should require revbump

[Matthew Thode]

[-- Attachment #1: Type: text/plain, Size: 680 bytes --]

On 10/19/2012 08:52 AM, Ian Stakenvicius wrote:
> Hey all -- there was just a discussion in #gentoo-dev about this, so
> following up here..
> 
> Due to the fact that an EAPI-bump can imply different behaviour from
> eclasses or the PM, an EAPI-bump should in most cases also require an
> ebuild revbump.
> 
> Why i'm bringing this up here, is because Chansaw and I were wondering
> if common sense will not be enough to ensure this and it should be
> made a policy to revbump when migrating to a new EAPI ?
> 
> 

Well, it makes sense that it requires it, but better to make it well
known then to rely on common sense.

-- 
-- Matthew Thode (prometheanfire)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [gentoo-project] EAPI bump should require revbump

[Dane Smith]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 10/19/12 09:52, Ian Stakenvicius wrote:
> Hey all -- there was just a discussion in #gentoo-dev about this,
> so following up here..
> 
> Due to the fact that an EAPI-bump can imply different behaviour
> from eclasses or the PM, an EAPI-bump should in most cases also
> require an ebuild revbump.
> 
> Why i'm bringing this up here, is because Chansaw and I were
> wondering if common sense will not be enough to ensure this and it
> should be made a policy to revbump when migrating to a new EAPI ?
> 
> 

- From where I sit, it just makes sense to revbump if you're changing
the EAPI. However, past experience tells me it often makes sense to
make things explicit. I don't like the idea of having tons and tons of
policies, but sometimes it helps if things are clear rather than implied.

- -- 
Dane Smith (c1pher)
Gentoo Linux Developer -- QA / Crypto
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
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=9gKs
-----END PGP SIGNATURE-----

Re: [gentoo-project] EAPI bump should require revbump

[Ulrich Mueller]

>>>>> On Fri, 19 Oct 2012, Ian Stakenvicius wrote:

> Hey all -- there was just a discussion in #gentoo-dev about this, so
> following up here..

> Due to the fact that an EAPI-bump can imply different behaviour from
> eclasses or the PM, an EAPI-bump should in most cases also require an
> ebuild revbump.

> Why i'm bringing this up here, is because Chansaw and I were wondering
> if common sense will not be enough to ensure this and it should be
> made a policy to revbump when migrating to a new EAPI ?

So far the guideline was that a revbump isn't required if the files
installed by the ebuild don't change, or if there are only trivial
changes that don't affect functionality (like files going to
/usr/share/doc).

I don't see why EAPI bumps should be handled differently from other
changes to the ebuild. If the installed files don't change, why would
one impose upon the user to recompile the package?

Ulrich

Re: [gentoo-project] EAPI bump should require revbump

[Samuli Suominen]

On 20/10/12 13:14, Ulrich Mueller wrote:
>>>>>> On Fri, 19 Oct 2012, Ian Stakenvicius wrote:
>
>> Hey all -- there was just a discussion in #gentoo-dev about this, so
>> following up here..
>
>> Due to the fact that an EAPI-bump can imply different behaviour from
>> eclasses or the PM, an EAPI-bump should in most cases also require an
>> ebuild revbump.
>
>> Why i'm bringing this up here, is because Chansaw and I were wondering
>> if common sense will not be enough to ensure this and it should be
>> made a policy to revbump when migrating to a new EAPI ?
>
> So far the guideline was that a revbump isn't required if the files
> installed by the ebuild don't change, or if there are only trivial
> changes that don't affect functionality (like files going to
> /usr/share/doc).
>
> I don't see why EAPI bumps should be handled differently from other
> changes to the ebuild. If the installed files don't change, why would
> one impose upon the user to recompile the package?
>
> Ulrich
>

+1.   PM's that can't handle EAPI bump without revbump are broken in my 
eyes.

If the content doesn't change, then revbumps are *annoying waste of CPU 
cycles*

- Samuli

Re: [gentoo-project] EAPI bump should require revbump

[Agostino Sarubbo]

[-- Attachment #1: Type: text/plain, Size: 240 bytes --]

On Saturday 20 October 2012 13:14:22 Samuli Suominen wrote:
> If the content doesn't change, then revbumps are *annoying waste of CPU
> cycles*
+1
-- 
Agostino Sarubbo / ago -at- gentoo.org
Gentoo/AMD64 Arch Security Liaison
GPG: 0x7CD2DC5D

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 490 bytes --]

Re: [gentoo-project] EAPI bump should require revbump

[Rich Freeman]

On Sat, Oct 20, 2012 at 6:14 AM, Ulrich Mueller <ulm@gentoo.org> wrote:
> I don't see why EAPI bumps should be handled differently from other
> changes to the ebuild. If the installed files don't change, why would
> one impose upon the user to recompile the package?

It would probably be best for this reason to wait until the package
actually does change to do an EAPI bump.

One thing that would concern me is testing - I don't like the idea of
stable packages changing in any way, since that bypasses the testing
cycle (both the formal cycle by the arch team, and the informal one by
being keyworded for testing for 30 days).

Also, the on-disk image really does have to not change.  An EAPI bump
might cause subtle changes that the maintainer might not be aware of.

All that said, I think you have a decent point.

Rich

Re: [gentoo-project] EAPI bump should require revbump

[Pacho Ramos]

[-- Attachment #1: Type: text/plain, Size: 1346 bytes --]

El sáb, 20-10-2012 a las 13:14 +0300, Samuli Suominen escribió:
> On 20/10/12 13:14, Ulrich Mueller wrote:
> >>>>>> On Fri, 19 Oct 2012, Ian Stakenvicius wrote:
> >
> >> Hey all -- there was just a discussion in #gentoo-dev about this, so
> >> following up here..
> >
> >> Due to the fact that an EAPI-bump can imply different behaviour from
> >> eclasses or the PM, an EAPI-bump should in most cases also require an
> >> ebuild revbump.
> >
> >> Why i'm bringing this up here, is because Chansaw and I were wondering
> >> if common sense will not be enough to ensure this and it should be
> >> made a policy to revbump when migrating to a new EAPI ?
> >
> > So far the guideline was that a revbump isn't required if the files
> > installed by the ebuild don't change, or if there are only trivial
> > changes that don't affect functionality (like files going to
> > /usr/share/doc).
> >
> > I don't see why EAPI bumps should be handled differently from other
> > changes to the ebuild. If the installed files don't change, why would
> > one impose upon the user to recompile the package?
> >
> > Ulrich
> >
> 
> +1.   PM's that can't handle EAPI bump without revbump are broken in my 
> eyes.
> 
> If the content doesn't change, then revbumps are *annoying waste of CPU 
> cycles*
> 
> - Samuli
> 
> 

+1

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-project] EAPI bump should require revbump

[Roy Bamford]

On 2012.10.19 14:52, Ian Stakenvicius wrote:
> Hey all -- there was just a discussion in #gentoo-dev about this, so
> following up here..
> 
> Due to the fact that an EAPI-bump can imply different behaviour from
> eclasses or the PM, an EAPI-bump should in most cases also require an
> ebuild revbump.
> 
> Why i'm bringing this up here, is because Chansaw and I were 
> wondering
> if common sense will not be enough to ensure this and it should be
> made a policy to revbump when migrating to a new EAPI ?
> 
> 

Common sense is much rarer than you might expect.
It should be explicit.

-- 
Regards,

Roy Bamford
(Neddyseagoon) a member of
elections
gentoo-ops
forum-mods
trustees

Re: [gentoo-project] EAPI bump should require revbump

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 1033 bytes --]

On 20.10.2012 14.25, Rich Freeman wrote:
> On Sat, Oct 20, 2012 at 6:14 AM, Ulrich Mueller <ulm@gentoo.org> wrote:
>> I don't see why EAPI bumps should be handled differently from other
>> changes to the ebuild. If the installed files don't change, why would
>> one impose upon the user to recompile the package?
> 
> It would probably be best for this reason to wait until the package
> actually does change to do an EAPI bump.
> 
> One thing that would concern me is testing - I don't like the idea of
> stable packages changing in any way, since that bypasses the testing
> cycle (both the formal cycle by the arch team, and the informal one by
> being keyworded for testing for 30 days).
> 

I don't think changing EAPI for stable packages is allowed. That means
we are only talking about testing packages for this thread and I think
there it's fine to follow the established rule already mentioned. If
developers break things continuously without testing, take up the issue
with QA.

Regards,
Petteri



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 902 bytes --]

Re: [gentoo-project] EAPI bump should require revbump

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 300 bytes --]

On 19.10.2012 16.52, Ian Stakenvicius wrote:
> Hey all -- there was just a discussion in #gentoo-dev about this, so
> following up here..
> 

Also please post these things to gentoo-dev mailing list next time. This
mailing list is for project wide non technical issues.

Regards,
Petteri


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 902 bytes --]

Re: [gentoo-project] EAPI bump should require revbump

[Ian Stakenvicius]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

On 25/10/12 08:41 AM, Petteri Räty wrote:
> On 19.10.2012 16.52, Ian Stakenvicius wrote:
>> Hey all -- there was just a discussion in #gentoo-dev about
>> this, so following up here..
>> 
> I don't think changing EAPI for stable packages is allowed. That 
> means we are only talking about testing packages for this thread 
> and I think there it's fine to follow the established rule already 
> mentioned. If developers break things continuously without
> testing, take up the issue with QA.


[ Merge! ]


 Also please post these things
> to gentoo-dev mailing list next time. This mailing list is for 
> project wide non technical issues.
> 

It was suggested to me to post here because #1) it's a policy
discussion rather than a technical one, and #2) it reaches a bigger
audience as many devs have tuned out to gentoo-dev@ ..

As for your previous statement -- I was not aware of a policy that
excludes EAPI bumps from occurring on stable packages.  This would
certainly suffice to alleviate the original concerns and probably be
more effective as a policy than always requiring revbump on EAPI change.

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iF4EAREIAAYFAlCJO7kACgkQ2ugaI38ACPBSzQD8Cj5o3NdNSRU7d/GmUwPw5nnl
whHrdX4s7kYjU7uMZz4A/izpnDcekv/QNaskk6LUHuUPnFNPcRMGc2nlY0qACZYY
=JHtD
-----END PGP SIGNATURE-----

Re: [gentoo-project] EAPI bump should require revbump

[Petteri Räty]

[-- Attachment #1: Type: text/plain, Size: 1375 bytes --]

On 25.10.2012 16:16, Ian Stakenvicius wrote:

> 
> 
>  Also please post these things
>> to gentoo-dev mailing list next time. This mailing list is for 
>> project wide non technical issues.
> 
> 
> It was suggested to me to post here because #1) it's a policy
> discussion rather than a technical one, and #2) it reaches a bigger
> audience as many devs have tuned out to gentoo-dev@ ..
> 

My educated guess is that most of those who are tuned out of gentoo-dev
don't follow this mailing list either. Last time we looked at
subscription numbers gentoo-project had less people than gentoo-dev.
Those who suggested you to post here didn't know our mailing list
policies. Granted it's not always easy to know where to post. Discussing
technical policies belongs to gentoo-dev. Especially in matters like
these that concern the Portage tree.

>
> As for your previous statement -- I was not aware of a policy that
> excludes EAPI bumps from occurring on stable packages.  This would
> certainly suffice to alleviate the original concerns and probably be
> more effective as a policy than always requiring revbump on EAPI change.
> 

There's a general policy that one should refrain from changing the logic
for ebuilds marked stable. EAPIs so far haven't introduced features that
would require being urgently but to stable.

regards,
Petteri


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