Re: FREEZE!

[aj@dungeon.inka.de (Andreas Jellinghaus)]

Guy wrote :
[...]
> Distribution non-free                   ->  hamm/non-free
> Distribution unstable, Section non-free ->  hamm/non-free (alias of above)
> Distribution contrib                    ->  hamm/contrib/
> Distribution unstable, Section contrib  ->  hamm/contrib/ (alias of above)
> Distribution unstable, Section *        ->  hamm/

a) we don't have a valid Distribution ("non-free" and "contrib" don't
say if a program belongs to unstable or stable), or
b) we don't have a valid Section ("non-free" or "contrib" is not a good
section IMO).

it's good, that we have seperated trees, but shouldn't we have 
a) a third control field, indicating mainstream/contrib/non-free or
b) Distribution Paths like "unstable/non-free" ?

this way we would have a valid section (mail/news/...), a valid
distribtion (unstable/stable/...) and the information if the package is
non-free or contrib. 
 
---
a different topic :

what happend to the plans of having a "tested" distribution ?
most developers agreed, that a program should not get into stable, yust
because it was in unstable for (up to) 3 months.

we can change theese thing now, so next release (2.0) will have theese
changes. we should not drop theese topics for the next 3 months, or
discuss them for 3 months.

regards andreas

> 
> 
> Unlike rex's freeze, uploads will not be allowed to automatically
> enter frozen.  All uploads to frozen will have to be manually
> confirmed.
> 
> I have improved dinstall quite a bit so if necessary, it rejects your
> file immediately even if it's new or going into frozen or stable.  So
> if dinstall appears to ignore it, it's actually processed it but is
> waiting on manual confirmation.  It will warn you in the initial
> processing if it sees something suspicous (for example frozen w/o
> unstable).
> 
> A final note.  I trust you all already know this, but only bug fixes
> may go into frozen.  No new packages, no major upstream changes.
> Also, listing frozen without listing unstable is almost definitely an
> error.
> 
> 
> Guy
> 
>