Re: Patch for smail
Christopher C Chimelis writes:
>
> On Mon, 1 Feb 1999, Bart Warmerdam wrote:
>
> > Brian rejected the i386 part (source and binary). The alpha part got in
> > though. The comment was "upload for alpha only". So I'll try to talk to the
> > maintener again... :)
>
> Yes and no. I would take it back up with Brian stating that we are a
> release arch candidate here and that the same rules should apply for us as
> they do with i386, basically that if the "whole source" cannot be
> distributed with the dist, then we can no longer claim that all of our
> source and patches are included and available. In short, if someone likes
> to "roll their own dist" with Debian on an Alpha, they would not have a
> smail that worked without unaligned traps everywhere. I could see it if we
> were an arch that wasn't releasing, but this is an important update to an
> important package. Once again, it seems that Alpha is being treated as a
> "second-class citizen"....
I think we cannot underestimate some technical issues, uploading new
source for an arch can introduce problems for another, which will need
a new upload, and this can creates long delays (if not an endless loop).
IMHO, we are to the point where we should not need to rebuilt a
package for every arch, because of a bug fix relevant to only one
(of course to be fair, this principle should apply to i386 also).
What may be done fo instance is :
- tolerate (or maybe enforce) that some arch have a binary package not
corresponding to the latest source in slink, if latest slink source has
a debian revision with 3 digit (the last digit means the last changes
was more or less arch-specific, or it was a 64bit issue).
Alternatively we could tolerate that an arch has some binary package
in slink from the source in potato, but that would be very awkward.
Loic
Reply to: