[gentoo-dev] Re: [gentoo-core] Keywords policy

[Ryan Hill]

Alec Warner wrote:
> On 3/10/08, Ryan Hill <dirtyepic@g.o> wrote:
>> Jeroen Roovers wrote:
>>  > On Mon, 10 Mar 2008 16:26:19 +0100
>>  > "Wulf C. Krueger" <philantrop@g.o> wrote:
>>  >
>>  >> No, we didn't because the whole thing is p.masked for a reason. It,
>>  >> KDE 4.0.1, is broken crap that should not yet be re-keyworded.
>>  >
>>  > OK then. and I am not going to cross-post this to -dev@, btw: why the
>>  > hell did you decide to put broken crap in the tree? It should never have
>>  > left your repository, it seems.
>>
>> It's package masked and unkeyworded, which is a big hint that it's under
>>  development.
>
> So Jer should just implicitly know not to keyword it?  Why not make it
> explicit?  That is all I am really asking for here.

How much more explicit can you make it than dropping every arch's keywords and 
putting it in package mask?  The problem here is that Jeroen decided that this 
was a violation of the keyword policy and blindly added his keywords back.  Fair 
enough, everyone makes a mistake from time to time.  But after more than a few 
people have tried to explain why this was a mistake, he still refuses to admit 
it and claims the keywords were dropped illegally.  I'm just pointing out that 
this is not the case, and never has been.  If a maintainer package masks an 
ebuild, you don't mess with it without talking to them.  This is coming straight 
from the handbook.

>>  > If you still wonder why I started rekeywording for HPPA, then let this
>>  > be the final answer. It was no fault of mine - I did it on purpose. No
>>  > keywording error - I was going to finish all the dependencies if you
>>  > hadn't asked me not to (because by then you were claiming KDE team
>>  > "reserves" the "right" to drop keywords at will and without notifying
>>  > arch teams, as opposed to current policy. The repoman bug / missing
>>  > feature left a few stones unturned, sadly, but I was going to do all of
>>  > KDE 4.
>>
>> You're still not getting this.  The KDE team did not _want_ these ebuilds
>>  keyworded.  That's why they _weren't_ keyworded.  That's why there was no bug
>>  filed, saying "hey we dropped these keywords" because they _did not want_ you to
>>  add them back yet.  When the ebuilds were of sufficient quality that they could
>>  be tested, then a bug is filed, the ebuilds are tested, and then re-keyworded.
>
> Right, but you did not make your want known, so how is Jer to know?

>>  Maintainers have every right to drop keywords if they think changes to their
>>  package are drastic enough to require re-evaluation by an architecture team.
>>  It's how we keep big fat calamity from befalling our users.  Yes, they need to
>>  inform the arch teams to re-add their keywords.  No that request does not need
>>  to come immediately if they're not ready for it.
>>
>>  A simple rule to go by:  Dropped keywords on package.masked packages are not
>>  dropped keywords.  If that package comes out of package.mask and still lacks
>>  your keyword and no bug is filed, then yes, then you have a legitimate beef.
>>
>>  This is simply the way things work from my point of view as a maintainer and a
>>  arch dev for a oft keyword-dropped arch.
>
> RIght but if everyone is not following the same rules you
> get...well...this exact situation.  The whole point of this discussion
> is not to assign blame, it is to figure out what we should change so
> this doesn't happen again as it obviously upset lots of folks.

As far as I know this is policy.  It has worked so far, but if something needs 
to change then so be it.


-- 
fonts, gcc-porting,                               by design, by neglect
mips, treecleaner,                        for a fact or just for effect
wxwidgets @ gentoo     EFFD 380E 047A 4B51 D2BD C64F 8AA8 8346 F9A4 0662