[gentoo-dev] Profile masking and profiles package.mask

[gentoo-dev] Profile masking and profiles package.mask

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 1892 bytes --]

This is a discussion to follow up bug #149508 [1].

The bug points to a behaviour change in handling of the profiles file, that, 
in my opinion at least, needs to be discussed, as there are profiles relying 
on the old behaviour (Gentoo/FreeBSD's to state some).

For what I can tell, the current behaviour has the advantage of providing a 
different masking reason for packages that are *needed to some version* for 
the profile to be complete, and for packages that are know not to work on a 
profile.

Example: Gentoo/FreeBSD relies on profiles masking for sys-freebsd/freebsd-* 
packages, as you should *not* use freebsd-lib 6.2 on the 6.1 profile, for 
instance; AMD64 no-multilib profiles use package.mask to mask packages that 
are known to be broken on that profile.

In case of Gentoo/FreeBSD, it also means to have 3x entries for forcing 
versions of the packages on users.

Another reason I'd see for retain the current behaviour is that users are 
known to unmask stuff via package.unmask to try "might-be-broken" versions. 
Considering that -* masking is deprecated, this means that if 2.4 profiles 
released a new version of linux-headers with some experimental support (okay, 
unlikely, but let's say it happens), it should go in package.mask.. user put 
linux-headers in package.unmask without a version (which is usually correct, 
as you might want to unmask newer revisions too), but find himself with 
linux-headers 2.6 unmasked.

I cannot find myself any reason for such a behaviour change, but I'm open to 
be proven wrong.


*Important: do NOT use this thread for considerations on QA behaviour, this is 
NOT what this post is thought for.*


[1] https://bugs.gentoo.org/show_bug.cgi?id=149508
-- 
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 384 bytes --]

On Sat, 30 Sep 2006 06:40:07 +0200 "Diego 'Flameeyes' Pettenò"
<flameeyes@gentoo.org> wrote:
| This is a discussion to follow up bug #149508 [1].

https://bugs.gentoo.org/show_bug.cgi?id=149536#c4

-- 
Ciaran McCreesh
Mail                : ciaranm at ciaranm.org
Web                 : http://ciaranm.org/
as-needed is broken : http://ciaranm.org/show_post.pl?post_id=13


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Jakub Moc]

[-- Attachment #1: Type: text/plain, Size: 1017 bytes --]

Ciaran McCreesh wrote:
> On Sat, 30 Sep 2006 06:40:07 +0200 "Diego 'Flameeyes' Pettenò"
> <flameeyes@gentoo.org> wrote:
> | This is a discussion to follow up bug #149508 [1].
> 
> https://bugs.gentoo.org/show_bug.cgi?id=149536#c4

If I were you, I'd rather not mention that bug. Really don't see what
you are trying to say here with that link. Plus kindly note

> *Important: do NOT use this thread for considerations on QA behaviour,
> this is NOT what this post is thought for.*

in the original mail. So now if you could move to the points mentioned
by Flameeyes in his email, it would be really useful. Additionally, it
would be nice if these discussions involved concerned arches and were
not done ex post in future cases.

Thanks.


-- 
Best regards,

 Jakub Moc
 mailto:jakub@gentoo.org
 GPG signature:
 http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
 Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95  B30F 8717 D5FD CEBA 3D9E

 ... still no signature   ;)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 563 bytes --]

On Saturday 30 September 2006 14:25, Ciaran McCreesh wrote:
> https://bugs.gentoo.org/show_bug.cgi?id=149536#c4
You bring up the point that you don't take any argument?

The argument is still valid, nobody provided a reason for the change.
I don't take anybody's word as a granted, so I don't care if Mike thinks the 
change is fine. I want some reasons.

If you want to waste even more time, continue this way.

-- 
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 879 bytes --]

On Sat, 30 Sep 2006 14:37:59 +0200 Jakub Moc <jakub@gentoo.org> wrote:
| Additionally, it would be nice if these discussions involved
| concerned arches and were not done ex post in future cases.

Uh, Jakub, part of the design of the devmanual was that it would be
possible for the right people to update it to codify existing practice
without arguments from the peanut gallery who like to claim that
because they've been getting away with it it's allowed. See also
conflicting USE flags and static metadata requirement... This isn't a
change, it's a documentation of how things are.

You've already wasted too much of other people's time on this. You're
not getting any more of mine.

-- 
Ciaran McCreesh
Mail                : ciaranm at ciaranm.org
Web                 : http://ciaranm.org/
as-needed is broken : http://ciaranm.org/show_post.pl?post_id=13


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Jakub Moc]

[-- Attachment #1: Type: text/plain, Size: 1407 bytes --]

Ciaran McCreesh wrote:
> On Sat, 30 Sep 2006 14:37:59 +0200 Jakub Moc <jakub@gentoo.org> wrote:
> | Additionally, it would be nice if these discussions involved
> | concerned arches and were not done ex post in future cases.
> 
> Uh, Jakub, part of the design of the devmanual was that it would be
> possible for the right people to update it to codify existing practice
> without arguments from the peanut gallery who like to claim that
> because they've been getting away with it it's allowed. See also
> conflicting USE flags and static metadata requirement... This isn't a
> change, it's a documentation of how things are.

You still didn't bring in any argument, why am I not surprised? Plus
don't pull in unrelated stuff into this debate. Plus getting devmanual
changed on the fly and using it as a point in discussion five minutes
later is not an acceptable practice for anyone who wants to debate
things seriously.

> You've already wasted too much of other people's time on this. You're
> not getting any more of mine.

Please, take this, your peanut galleries and other junk off-list,
noone's interested.

Thanks.

-- 
Best regards,

 Jakub Moc
 mailto:jakub@gentoo.org
 GPG signature:
 http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
 Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95  B30F 8717 D5FD CEBA 3D9E

 ... still no signature   ;)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 828 bytes --]

On Sat, 30 Sep 2006 14:40:44 +0200 "Diego 'Flameeyes' Pettenò"
<flameeyes@gentoo.org> wrote:
| On Saturday 30 September 2006 14:25, Ciaran McCreesh wrote:
| > https://bugs.gentoo.org/show_bug.cgi?id=149536#c4
|
| You bring up the point that you don't take any argument?
| 
| The argument is still valid, nobody provided a reason for the change.

It is not a change in policy. It's a codification of existing practice.
Unfortunately, since most of this stuff didn't get written up years ago
when it happened, not everyone is aware of said practice and so a few
profiles from developers who weren't around at the time don't honour it.

-- 
Ciaran McCreesh
Mail                : ciaranm at ciaranm.org
Web                 : http://ciaranm.org/
as-needed is broken : http://ciaranm.org/show_post.pl?post_id=13


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 1046 bytes --]

On Saturday 30 September 2006 15:14, Ciaran McCreesh wrote:
> It is not a change in policy. It's a codification of existing practice.
The behaviour of portage seems to ask you to differ on this. But you also seem 
to lose your point.

I'm discussing the change of behaviour with respect to portage, not the change 
to devmanual. Why am I not questioning the change to devmanual? Because the 
bug there is just required by Plasmaroo who act as a proxy between QA and 
devmanual itself, and thus not up to discuss anymore than a CVS commit.

But, the behaviour change for profiles (that currently, for bug or design, 
acts in a different way than your "existing practice" suggest) is up to 
debate, as I've found a few reasons not to back up such a change, and rather 
stick with the current behaviour.

So, again, why should we change the behaviour, other than to fullfill your 
request?

-- 
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Marius Mauch]

On Sat, 30 Sep 2006 06:40:07 +0200
"Diego 'Flameeyes' Pettenò" <flameeyes@gentoo.org> wrote:

> This is a discussion to follow up bug #149508 [1].
> 
> The bug points to a behaviour change in handling of the profiles
> file, that, in my opinion at least, needs to be discussed, as there
> are profiles relying on the old behaviour (Gentoo/FreeBSD's to state
> some).
> 
> For what I can tell, the current behaviour has the advantage of
> providing a different masking reason for packages that are *needed to
> some version* for the profile to be complete, and for packages that
> are know not to work on a profile.

[snip]

Personally I dislike the masking aspect of the packages file, as it's
mostly redundant, problematic in some cases (e.g. requring a
specific gcc versions masks all older gcc versions implicitly) and I
think having a single file to serve two purposes (set "system" and
masking packages) is crappy. Also overriding profile masks (yes,
this is valid sometimes) isn't intuitive either as there is no
"unmask" feature. This isn't connected to the mentioned bug at all btw.

However I understand your reasoning about unmasking things with
package.unmask, the question is how common that use case would be?

Marius

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 152 bytes --]

seriously jakub, stop responding ... you have nothing technical to offer to 
the issue at hand

let the people who work on portage handle it
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Jakub Moc]

[-- Attachment #1: Type: text/plain, Size: 1141 bytes --]

Mike Frysinger wrote:
> seriously jakub, stop responding ... you have nothing technical to offer to 
> the issue at hand
> 
> let the people who work on portage handle it
> -mike

Eh, the whole technical point here is that paludis behaviour differs
from portage (and differs from pkgcore, FWIW).

So, hiding the inconsistency via altering the profiles doesn't change
anything. Plus, the point of the bug's flame fest was that bugzilla is
not a proper place to request such behaviour changes, and definitely not
a reason for QA to mess with the profiles. Sticking the stuff in
package.mask won't make the inconsistent behaviour vanish in any way, it
will just hide it.

So, I'd kinda appreciate if concerned folks (including portage and
relevant affected arches) were involved in this discussion, instead of
sneaking the changes in under QA disguise.

Thanks.


-- 
Best regards,

 Jakub Moc
 mailto:jakub@gentoo.org
 GPG signature:
 http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
 Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95  B30F 8717 D5FD CEBA 3D9E

 ... still no signature   ;)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 458 bytes --]

On Saturday 30 September 2006 13:02, Jakub Moc wrote:
> Eh, the whole technical point here is that paludis behaviour differs
> from portage (and differs from pkgcore, FWIW).

the technical point is what is the expected behavior of the packages file ... 
seems silly to duplicate masking across two different files

what have you offered to this discussion ?  nothing: so sit back and let the 
people who actually work on this stuff handle it
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Jakub Moc]

[-- Attachment #1: Type: text/plain, Size: 1699 bytes --]

Mike Frysinger wrote:
> On Saturday 30 September 2006 13:02, Jakub Moc wrote:
>> Eh, the whole technical point here is that paludis behaviour differs
>> from portage (and differs from pkgcore, FWIW).
> 
> the technical point is what is the expected behavior of the packages file ... 
> seems silly to duplicate masking across two different files
> 
> what have you offered to this discussion ?  nothing: so sit back and let the 
> people who actually work on this stuff handle it
> -mike

It's not duplicating, exactly the opposite. Sticking the stuff into
per-profile package.mask is duplicating the information, because portage
 handles it just fine without any such duplication (that's the whole
point of Flameeyes' original mail).

Now if you want to change this, nothing wrong with that except when
someone goes moaning to bugzilla and QA starts messing with the profiles
without any discussion. This is not a QA issue.

If you want to change this behaviour, go provide some reason why it
should be done and either you persuade the folks involved or not. In
addition, those two kinds of masking (masked by profile vs. masked by
package.mask) are not duplicating each other, they behave quite differently.

For the rest, sorry but replies as "stop wasting everyone's time" or
"sit back and let people who actually work do it" as not much polite and
don't offer anything to this discussion either.

Thanks.

-- 
Best regards,

 Jakub Moc
 mailto:jakub@gentoo.org
 GPG signature:
 http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
 Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95  B30F 8717 D5FD CEBA 3D9E

 ... still no signature   ;)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 1643 bytes --]

On Saturday 30 September 2006 00:40, Diego 'Flameeyes' Pettenò wrote:
> For what I can tell, the current behaviour has the advantage of providing a
> different masking reason for packages that are *needed to some version* for
> the profile to be complete, and for packages that are know not to work on a
> profile.

isnt that the point of putting a comment above a mask ?
# this package wont work on this profile
bar/foo
# these versions are needed in this profile
=cat/meow-1*

> Example: Gentoo/FreeBSD relies on profiles masking for
> sys-freebsd/freebsd-* packages, as you should *not* use freebsd-lib 6.2 on
> the 6.1 profile, for instance; AMD64 no-multilib profiles use package.mask
> to mask packages that are known to be broken on that profile.
>
> In case of Gentoo/FreeBSD, it also means to have 3x entries for forcing
> versions of the packages on users.

i dont get it ... if you mask the versions in package.mask, why do you need a 
packages entry at all ?

fbsd/packages:sys-freebsd/freebsd-mk-defs
fbsd/package.mask:<nothing>
fbsd/6.1/packages:<nothing>
fbsd/6.1/package.mask:>=sys-freebsd/freebsd-mk-defs-6.2
fbsd/6.2/packages:<nothing>
fbsd/6.2/package.mask:<nothing>

> Another reason I'd see for retain the current behaviour is that users are
> known to unmask stuff via package.unmask to try "might-be-broken" versions.

so what you're arguing is that we should retain the existing behavior because 
users might try to unmask something properly ?  trying to protect users from 
shooting themselves in the foot by making the profile behavior more 
complicated is a waste of time
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Danny van Dyk]

Am Samstag, 30. September 2006 19:02 schrieb Jakub Moc:
> Mike Frysinger wrote:
> > seriously jakub, stop responding ... you have nothing technical to
> > offer to the issue at hand
> >
> > let the people who work on portage handle it
> > -mike
>
> Eh, the whole technical point here is that paludis behaviour differs
> from portage (and differs from pkgcore, FWIW).
This has little to do with why this change to the devmanual has been 
done.

> So, hiding the inconsistency via altering the profiles doesn't change
> anything. Plus, the point of the bug's flame fest was that bugzilla
> is not a proper place to request such behaviour changes, and
> definitely not a reason for QA to mess with the profiles. Sticking
> the stuff in package.mask won't make the inconsistent behaviour
> vanish in any way, it will just hide it.
It is not a behaviour change imho. The "packages" file changed
its meaning subtly after introducing cascading profiles.
As ciaranm already pointed out: It is not meant to mask "<"-like 
versions anymore. It's meant to
- Describe the system package set
- Define which versions are _at least_ needed for a profile.

> So, I'd kinda appreciate if concerned folks (including portage and
> relevant affected arches) were involved in this discussion, instead
> of sneaking the changes in under QA disguise.
Release engineering arch coordinators, which happen to be the people who
maintain the profiles below default-linux/ for their relevant arches, 
have been CCed and Chris already stated that he forgot/didn't realize
to fix this problem for no-nptl/2.4's package.mask.

Jakub: Please reevaluate the behaviour you showed on both the bug and 
this mailing list. I for one don't consider it anywhere near 
appropriate. This shall be no offense, just a comment in regard that 
you can do better.

Danny
-- 
Danny van Dyk <kugelfang@gentoo.org>
Gentoo/AMD64 Project, Gentoo Scientific Project
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Jochen Maes]

Danny van Dyk wrote:
> Am Samstag, 30. September 2006 19:02 schrieb Jakub Moc:
>   
>> Mike Frysinger wrote:
>>     
>>> seriously jakub, stop responding ... you have nothing technical to
>>> offer to the issue at hand
>>>
>>> let the people who work on portage handle it
>>> -mike
>>>       
>> Eh, the whole technical point here is that paludis behaviour differs
>> from portage (and differs from pkgcore, FWIW).
>>     
> This has little to do with why this change to the devmanual has been 
> done.
>
>   
>> So, hiding the inconsistency via altering the profiles doesn't change
>> anything. Plus, the point of the bug's flame fest was that bugzilla
>> is not a proper place to request such behaviour changes, and
>> definitely not a reason for QA to mess with the profiles. Sticking
>> the stuff in package.mask won't make the inconsistent behaviour
>> vanish in any way, it will just hide it.
>>     
> It is not a behaviour change imho. The "packages" file changed
> its meaning subtly after introducing cascading profiles.
> As ciaranm already pointed out: It is not meant to mask "<"-like 
> versions anymore. It's meant to
> - Describe the system package set
> - Define which versions are _at least_ needed for a profile.
>
>   
>> So, I'd kinda appreciate if concerned folks (including portage and
>> relevant affected arches) were involved in this discussion, instead
>> of sneaking the changes in under QA disguise.
>>     
> Release engineering arch coordinators, which happen to be the people who
> maintain the profiles below default-linux/ for their relevant arches, 
> have been CCed and Chris already stated that he forgot/didn't realize
> to fix this problem for no-nptl/2.4's package.mask.
>
> Jakub: Please reevaluate the behaviour you showed on both the bug and 
> this mailing list. I for one don't consider it anywhere near 
> appropriate. This shall be no offense, just a comment in regard that 
> you can do better.
>   
mike, danny,
thanks for trying, but past reference showed that he likes to talk like
a chicken who's head has been chopped of.
This whole discussion made most of the people forget what it was about...
good on ya jakub...
> Danny
>   

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Profile masking and profiles package.mask

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 1684 bytes --]

On Saturday 30 September 2006 19:39, Mike Frysinger wrote:
> isnt that the point of putting a comment above a mask ?
> # this package wont work on this profile
> bar/foo
Indeed, but the problem is that the masks are all normalised in one big mask. 
Which means that users might want to unmask certain versions found in the 
top-level profile.mask, and also unmask some of the packages masked in a 
profile.

> fbsd/packages:sys-freebsd/freebsd-mk-defs
> fbsd/package.mask:<nothing>
> fbsd/6.1/packages:<nothing>
> fbsd/6.1/package.mask:>=sys-freebsd/freebsd-mk-defs-6.2
> fbsd/6.2/packages:<nothing>
> fbsd/6.2/package.mask:<nothing>
Actually, you need to mask < versions, too ...

> so what you're arguing is that we should retain the existing behavior
> because users might try to unmask something properly ?  trying to protect
> users from shooting themselves in the foot by making the profile behavior
> more complicated is a waste of time
Uh, it's not "making the profile behaviour more complicated", it's "retaining 
the current behaviour of profiles".

But seems I'm in minority on this.
Still, if we're going to change this behaviour, it's the case to do it 
properly, by also updating the behaviour of portage itself, and document this 
properly (as in, give a reasoning for this change of behaviour).

Note to Danny: releng controls default-linux, okay, but there are other 
profiles than those, hardened and Gentoo/Alt. The decision should have been 
taken by all the three of us, not unilaterally.

-- 
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Mike Frysinger]

[-- Attachment #1: Type: text/plain, Size: 1374 bytes --]

On Saturday 30 September 2006 20:06, Diego 'Flameeyes' Pettenò wrote:
> On Saturday 30 September 2006 19:39, Mike Frysinger wrote:
> > isnt that the point of putting a comment above a mask ?
> > # this package wont work on this profile
> > bar/foo
>
> Indeed, but the problem is that the masks are all normalised in one big
> mask. Which means that users might want to unmask certain versions found in
> the top-level profile.mask, and also unmask some of the packages masked in
> a profile.

i dont understand what you're trying to say here ... the behavior you're 
describing sounds correct to me ... provide some examples ?

> > fbsd/packages:sys-freebsd/freebsd-mk-defs
> > fbsd/package.mask:<nothing>
> > fbsd/6.1/packages:<nothing>
> > fbsd/6.1/package.mask:>=sys-freebsd/freebsd-mk-defs-6.2
> > fbsd/6.2/packages:<nothing>
> > fbsd/6.2/package.mask:<nothing>
>
> Actually, you need to mask < versions, too ...

sure ... not that people should be downgrading in the first place (glibc 
ebuilds prevent this), but you are correct

> Note to Danny: releng controls default-linux, okay, but there are other
> profiles than those, hardened and Gentoo/Alt. The decision should have been
> taken by all the three of us, not unilaterally.

there is no central body for profiles ... and more projects than just these 
three are affected
-mike

[-- Attachment #2: Type: application/pgp-signature, Size: 827 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 1960 bytes --]

On Sat, 2006-09-30 at 19:02 +0200, Jakub Moc wrote:
> So, I'd kinda appreciate if concerned folks (including portage and
> relevant affected arches) were involved in this discussion, instead of
> sneaking the changes in under QA disguise.

Umm... I already took care of x86/alpha, both of which I am on the arch
team for, as well as added release@ to CC, since most of the profiles
involved were made by Release Engineering and the architecture teams.
There were a few people that I missed in the discussions, namely
Gentoo/*BSD.  This was not on purpose but more an oversight.  I tried to
kindly ask you the other day on IRC to add items to the bug that needed
to be done, rather than continue with your current course of action of
trying to lay blame to everyone when they make a mistake.  At this
point, I'll have to agree with Mike by saying that if you're not going
to assist on a technical level with the bug and discussion, to please
not comment, at all.

Diego, a few people from the portage team have said that they dislike
using version masks in the packages file.  I agree with them completely.
It causes quite a few problems.  To be simple, all masks should be in
package.mask rather than in packages.  This has been the case since the
per-profile package.mask was introduced.  Unfortunately, as with many
things in the portage world, we can't just turn off "old" functionality
over night due to the amount of system that we would break.  This means
there needs to be a transition period.  The fact that something works
now really is more of a side-effect than an expected behavior.

I plan on going through all of the "default-linux" profiles and cleaning
them up in this regard, after speaking with each arch team that seems
affected by the change.

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Jakub Moc]

[-- Attachment #1: Type: text/plain, Size: 973 bytes --]

Chris Gianelloni wrote:

> I tried to
> kindly ask you the other day on IRC to add items to the bug that needed
> to be done, rather than continue with your current course of action of
> trying to lay blame to everyone when they make a mistake.  At this
> point, I'll have to agree with Mike by saying that if you're not going
> to assist on a technical level with the bug and discussion, to please
> not comment, at all.

Sure, and I did tell you about those that I have found, such as
uclibc/*/2.4 and default-linux/sparc/sparc{32,64}/2006.1/2.4 [1], also
told you that I don't have time to ponder thru all the profiles in there.

[1] http://bugs.gentoo.org/show_bug.cgi?id=149508#c26

Sigh...


-- 
Best regards,

 Jakub Moc
 mailto:jakub@gentoo.org
 GPG signature:
 http://subkeys.pgp.net:11371/pks/lookup?op=get&search=0xCEBA3D9E
 Primary key fingerprint: D2D7 933C 9BA1 C95B 2C95  B30F 8717 D5FD CEBA 3D9E

 ... still no signature   ;)


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Diego 'Flameeyes' Pettenò]

[-- Attachment #1: Type: text/plain, Size: 1173 bytes --]

On Monday 02 October 2006 14:34, Chris Gianelloni wrote:
> Diego, a few people from the portage team have said that they dislike
> using version masks in the packages file.  I agree with them completely.
I've already seen I'm in a minority, by liking the current behaviour.

> Unfortunately, as with many 
> things in the portage world, we can't just turn off "old" functionality
> over night due to the amount of system that we would break.  This means
> there needs to be a transition period.
Of course, and that also needs to be properly documented and planned, or we'll 
be facing the same issues in a while again.

I tried to bring the discussion about this a) because I prefer the current 
behaviour b) because at least it's known what's going on.
Seems like I'm the only one liking the current behaviour, so I gladly shut up, 
just wanted to be sure if this was the case (and having something to quote if 
something breaks when the behaviour is switched ;) happens too often not to 
consider this, too).

-- 
Diego "Flameeyes" Pettenò - http://farragut.flameeyes.is-a-geek.org/
Gentoo/Alt lead, Gentoo/FreeBSD, Video, AMD64, Sound, PAM, KDE

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] OT noise (Was: Profile masking and profiles package.mask)

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 847 bytes --]

On Mon, 2006-10-02 at 14:50 +0200, Diego 'Flameeyes' Pettenò wrote:
> just wanted to be sure if this was the case (and having something to quote if 
> something breaks when the behaviour is switched ;) happens too often not to 
> consider this, too).

Yeah.  It would be nice if we had some good way of getting this sort of
information out to the developers.  All too often we have behavior
changes that people don't know about.  I know that it has been suggested
a few times before, but what do people think about the possibility of
some sort of "-dev-announce" list or something to put useful
non-discussion information such as this, with reply-to set to this list?

-- 
Chris Gianelloni
Release Engineering Strategic Lead
Alpha/AMD64/x86 Architecture Teams
Games Developer/Council Member/Foundation Trustee
Gentoo Foundation

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Profile masking and profiles package.mask

[Jason Stubbs]

On Saturday 30 September 2006 04:40, Diego 'Flameeyes' Pettenò wrote:
> This is a discussion to follow up bug #149508 [1].

Posted on the bug before noticing there was a -dev thread.

"""
Just about everybody has the wrong idea here.

1) Specifying <sys-libs/glibc-2.4 in packages *does* mask >=sys-libs/glibc-2.4
and thus a corresponding entry in package.mask

2) What should be done is to specify >=sys-libs/glibc-2.4 and leave masking 
out altogether for packages

The reason that package.mask was added to profiles was so that masking of 
atoms in packages could be killed off and it could become just a list of 
required packages.
"""

Like Marius said, using packages to both define what's required of "system" 
and for masking packages is bad design. That and the hope of eventually being 
able to kill off profiles/package.mask are the only reasons package.mask was 
introduced into profiles.

<snip stuff that Mike responded to correctly>

> I cannot find myself any reason for such a behaviour change, but I'm open
> to be proven wrong.

The original reason for specifying masking in both packages and package.mask 
was that there were portage versions that didn't look at package.mask. That 
was a long time ago though, so masking should really be dropped from packages 
altogether at this late stage.

However, masking in packages only is still supported. If there is a reason 
that the plans for killing off that support should be suspended, that's also 
viable.

--
Jason Stubbs

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Profile masking and profiles package.mask

[Jason Stubbs]

On Monday 02 October 2006 16:03, Jason Stubbs wrote:
> 1) Specifying <sys-libs/glibc-2.4 in packages *does* mask
> >=sys-libs/glibc-2.4 and thus a corresponding entry in package.mask

... is redundant

--
Jason Stubbs
-- 
gentoo-dev@gentoo.org mailing list