[gentoo-dev] Gentoo "stable" going in wrong direction ?

[gentoo-dev] Gentoo "stable" going in wrong direction ?

[Rainer Groesslinger]

Hello,

I have been looking around for sime time in various ebuilds since Gentoo
introduced the "stable" and "testing/unstable" feature...

Well, in general I think it is a very good idea but I don't think it is used
as it should be !

For example Maik "blizzy" Schreiber told me about
http://gentoo-stable.iq-computing.de which is something like a "voting
system" but almost nobody is using it (for example mozilla 1.2.1 only has
one vote although many thousand people are using it - with success) and if
you take a look at the ebuild you see that every mozilla ebuild with version
1.2.1 has the keyword ~x86 - so stable users don't get it although there's
no reason for calling Mozilla 1.2.1 "unstable"...

In my opinion http://gentoo-stable.iq-computing.de should be a more-or-less
official voting system for the packages or gentoo stable will end like
debian stable and I don't think Gentoo wants to go *that* stable :)

There are just not enough users and feedback pushing unstable packages to
stable from what I see...

There was/is talk about package.mask being removed in the future - good idea
but I think it should look like this

stable: KDE 3.0.5
unstable: KDE3.1RC5

stable: Mozilla 1.2.1
unstable: Mozilla 1.3a

and so on...In short: Gentoo stable should be as close as possible to what
the developers of the various applications call "stable" - why not believe
them ? ;p

Currently the package.mask carries packages which have a right to be called
unstable, e.g. XFree 4.2.99 and so on...
But the stable/unstable situation in some ebuilds is a bit confusing and
leading in the wrong direction if continued like this ?

Of course every distribution needs to test individual things, make some
changes here and there...And to avoid a bad stable tree I highly suggesst
using blizzy's system...

I didn't know of it, he just told me some minutes ago and I think it's a
great chance for people to vote for it etc. - if it's used and developers
set their ebuilds as 'stable' according to what people voted (or not)...


Just my opinion about current stable/unstable things...
Rainer



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

> I read your post and i had an idea, why not use the gentoo-stats to see 
> what people has installed on theyr machines and then if a lot of people 
> is installing the same stuff call that stuff "stable". I know there 
> should be some people always testing new software to make the usage 
> number increase and be promoted to the stable level, but this is going 
> to happen anyway.

If you mean marking ebuilds as "stable" automatically, I don't think that's such a good idea, for obvious reasons. However, it's possible
to automatically add "merged here" counters for those ebuilds in Gentoo Stable.

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Saverio Vigni]

Rainer Groesslinger wrote:

>Hello,
>
>I have been looking around for sime time in various ebuilds since Gentoo
>introduced the "stable" and "testing/unstable" feature...
>  
>
I read your post and i had an idea, why not use the gentoo-stats to see 
what people has installed on theyr machines and then if a lot of people 
is installing the same stuff call that stuff "stable". I know there 
should be some people always testing new software to make the usage 
number increase and be promoted to the stable level, but this is going 
to happen anyway.

Saverio Vigni
www.hor-net.com


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[foser]

Rainer Groesslinger wrote:
> For example Maik "blizzy" Schreiber told me about
> http://gentoo-stable.iq-computing.de which is something like a "voting
> system" but almost nobody is using it (for example mozilla 1.2.1 only has
> one vote although many thousand people are using it - with success) and if
> you take a look at the ebuild you see that every mozilla ebuild with version
> 1.2.1 has the keyword ~x86 - so stable users don't get it although there's
> no reason for calling Mozilla 1.2.1 "unstable"...

It depends -possibly- on the xft-2 ebuild and that isn't marked stable 
yet, that's the reason.

- foser


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

>>version
>>1.2.1 has the keyword ~x86 - so stable users don't get it although there's
>>no reason for calling Mozilla 1.2.1 "unstable"...
> 
> It depends -possibly- on the xft-2 ebuild and that isn't marked stable 
> yet, that's the reason.

If mozilla-1.2.1 is "unstable" because of _this_, I think that's an improper use of masking. There's really no point in masking something
because its _dependencies_ are masked. If you do this, you would have to check each dependency, and if all of them are "stable", you can
mask the package itself "stable" as well.

Instead, Portage should respect the "unstable" dependencies, and warn you that it can't install your "stable" package because some of its
dependencies are "unstable". (Portage already does that, which is good.)

So again, there's no point in masking a package "unstable" just because its dependencies are "unstable".

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Jon Portnoy]

On Mon, 16 Dec 2002, Maik Schreiber wrote:

[snip]
> 
> So again, there's no point in masking a package "unstable" just because its dependencies are "unstable".
> 
> -- 
> Maik Schreiber, Gentoo Linux Developer
> http://www.gentoo.org
> mailto:blizzy@gentoo.org
> 
> --
> gentoo-dev@gentoo.org mailing list
> 


Er? If the dependencies are marked unstable and the package isn't, nobody 
(or nobody only using stable keywords, I should say) can install it unless 
the dependencies are marked stable...

--
Jon Portnoy


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

>> So again, there's no point in masking a package "unstable" just because its dependencies are "unstable".
> 
> Er? If the dependencies are marked unstable and the package isn't, nobody 
> (or nobody only using stable keywords, I should say) can install it unless 
> the dependencies are marked stable...

Exactly, that's what is intended. So you simply don't gain anything with marking the package itself "unstable" as well, except for more
work.

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Matthew Walker]

Jon Portnoy said:
>
> On Mon, 16 Dec 2002, Maik Schreiber wrote:
>
> [snip]
>>
>> So again, there's no point in masking a package "unstable" just because
>> its dependencies are "unstable".
>>
>> --
>> Maik Schreiber, Gentoo Linux Developer
>> http://www.gentoo.org
>> mailto:blizzy@gentoo.org
>>
>> --
>> gentoo-dev@gentoo.org mailing list
>>
>
>
> Er? If the dependencies are marked unstable and the package isn't, nobody
> (or nobody only using stable keywords, I should say) can install it unless
>  the dependencies are marked stable...
>

True. But as soon as the dependencies get marked stable, they /will/ be able
to install it.

Also, in this case, the unstable dependency was purely an optional dependency.

> --
> Jon Portnoy
>
>
> --
> gentoo-dev@gentoo.org mailing list




--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Martin Schlemmer]

On Mon, 2002-12-16 at 22:34, Maik Schreiber wrote:
> >>version
> >>1.2.1 has the keyword ~x86 - so stable users don't get it although there's
> >>no reason for calling Mozilla 1.2.1 "unstable"...
> > 
> > It depends -possibly- on the xft-2 ebuild and that isn't marked stable 
> > yet, that's the reason.
> 
> If mozilla-1.2.1 is "unstable" because of _this_, I think that's an improper use of masking. There's really no point in masking something
> because its _dependencies_ are masked. If you do this, you would have to check each dependency, and if all of them are "stable", you can
> mask the package itself "stable" as well.
> 
> Instead, Portage should respect the "unstable" dependencies, and warn you that it can't install your "stable" package because some of its
> dependencies are "unstable". (Portage already does that, which is good.)
> 
> So again, there's no point in masking a package "unstable" just because its dependencies are "unstable".

Mozilla 1.2.1 is marked as testing, cause there are still some minor
issues with it.  Then, if anybody did mind checking, it do not
depend on x11-libs/xft, but compile it internally.  I have taken
much time to have its Xft2.0 'contained' ... you will see that its
not even libXft.so or libXrender.so anymore, but libXft_moz.so and
libXrender_moz.so ...  This change seems to fix some issues that
some  people had with it not starting, so it may be marked stable
in a bit if all goes well ...


-- 

Martin Schlemmer
Gentoo Linux Developer, Desktop/System Team Developer
Cape Town, South Africa



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

>> Er? If the dependencies are marked unstable and the package isn't, nobody
>> (or nobody only using stable keywords, I should say) can install it unless
>>  the dependencies are marked stable...
>>
> 
> True. But as soon as the dependencies get marked stable, they /will/ be able
> to install it.

Yes, and purely automagically. No work on the package itself would need to be done (ie. marking it as "stable").

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

> Mozilla 1.2.1 is marked as testing, cause there are still some minor
> issues with it.  Then, if anybody did mind checking, it do not
> depend on x11-libs/xft, but compile it internally.  I have taken
> much time to have its Xft2.0 'contained' ... you will see that its
> not even libXft.so or libXrender.so anymore, but libXft_moz.so and
> libXrender_moz.so ...  This change seems to fix some issues that
> some  people had with it not starting, so it may be marked stable
> in a bit if all goes well ...

Sorry if this was causing confusion with you, but Mozilla was just an example. The discussion was about marking an otherwise stable package
as "unstable" because some of its dependencies were marked "unstable".

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Martin Schlemmer]

On Mon, 2002-12-16 at 23:04, Maik Schreiber wrote:

> Sorry if this was causing confusion with you, but Mozilla was just an example. The discussion was about marking an otherwise stable package
> as "unstable" because some of its dependencies were marked "unstable".

NP =)


-- 

Martin Schlemmer
Gentoo Linux Developer, Desktop/System Team Developer
Cape Town, South Africa



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Matthew Walker]

foser said:
>
> And i'm not sure it's a good idea to stabilize top packages while some  of
> it's -needed- deps are still masked testing, i mean this will break
> things like -u world won't it ?

Yes, it would, currently. However, this is just an indication that portage
needs to skip updates that it can not satisfy the dependencies of, and
continue with the update instead of just dying.


Matthew
-- 
 Was I helpful?  Let others know:
 http://svcs.affero.net/rm.php?r=utoxin&p=main



--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Matthew Walker]

foser said:
> Matthew Walker wrote:
>> Yes, it would, currently. However, this is just an indication that
>> portage needs to skip updates that it can not satisfy the dependencies
>> of, and continue with the update instead of just dying.
>
> Just to get this clear, does or doesn't it break nowadays? I don't do
> much upgrade worlds anymore, so i can't tell ;)
>

It still breaks.

> - foser
>
>
> --
> gentoo-dev@gentoo.org mailing list




--
gentoo-dev@gentoo.org mailing list

[gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[foser]

Maik Schreiber wrote:
> If mozilla-1.2.1 is "unstable" because of _this_, I think that's an improper use of masking. There's really no point in masking something
> because its _dependencies_ are masked. If you do this, you would have to check each dependency, and if all of them are "stable", you can
> mask the package itself "stable" as well.
> 
> Instead, Portage should respect the "unstable" dependencies, and warn you that it can't install your "stable" package because some of its
> dependencies are "unstable". (Portage already does that, which is good.)
> 
> So again, there's no point in masking a package "unstable" just because its dependencies are "unstable".

Hm, got it wrong and mozilla doesnt use the system xft anymore (this 
saves some trouble i agree with Az, but it has it's disadvantages as 
well). Anyway, altough most users see a pack as stable there may still 
be trouble, afaics here Azarah was still tweaking it, so it should be 
masked and if it was intertwined with xft-2 then it surely should be 
cause that's an extra unstable factor.

And i'm not sure it's a good idea to stabilize top packages while some 
of it's -needed- deps are still masked testing, i mean this will break 
things like -u world won't it ?

- foser

PS. (devs only) Is all this crossposting necessary? Just keep this in 
-dev only.


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[foser]

Matthew Walker wrote:
> Yes, it would, currently. However, this is just an indication that portage
> needs to skip updates that it can not satisfy the dependencies of, and
> continue with the update instead of just dying.

Just to get this clear, does or doesn't it break nowadays? I don't do 
much upgrade worlds anymore, so i can't tell ;)

- foser


--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[Maik Schreiber]

> And i'm not sure it's a good idea to stabilize top packages while some 
> of it's -needed- deps are still masked testing, i mean this will break 
> things like -u world won't it ?

Yes, but this is a problem in Portage (emerge). It really shouldn't break.

> PS. (devs only) Is all this crossposting necessary?

In my opinion, I think it is.

-- 
Maik Schreiber, Gentoo Linux Developer
http://www.gentoo.org
mailto:blizzy@gentoo.org

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: [gentoo-core] Re: [gentoo-dev] Gentoo "stable" going in wrong direction ?

[foser]

Maik Schreiber wrote:
>>And i'm not sure it's a good idea to stabilize top packages while some 
>>of it's -needed- deps are still masked testing, i mean this will break 
>>things like -u world won't it ?
> 
> 
> Yes, but this is a problem in Portage (emerge). It really shouldn't break.

So, we're talking features not in portage yet. As it is imho it is 
better to mask top packages with masked deps.

- foser


--
gentoo-dev@gentoo.org mailing list