[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Using buildds only (was: Results of the meeting...)

* Wouter Verhelst (wouter@grep.be) wrote:
> On Tue, Aug 23, 2005 at 09:14:28AM -0400, Stephen Frost wrote:
> > Sure we do, for certain ports (ie: amd64).  Really, this just means it'd
> > be better to implement a system along the lines of:
> > 
> > source upload
> > fastest/preferred buildd type (i386, amd64, whatever) attempts build
> >  --> Success
> >     Other buildds attempt to build
> That would introduce some delay which, though generally probably not
> even noticeable, would become a problem if there's an issue with the
> "preferred" buildd (such as things being broken and the maintainer on a
> day trip, which isn't all that uncommon)

I wasn't thinking a specific/single buildd, but a "preferred"
architecture, which would have multiple buildds to hopefully alliviate
problems with one going down.  An interesting alternative would be to
have two sets of archs- tier-1 and tier-2, say, where tier-1's build
right away and tier-2's don't try unless some tier-1 succeeded (or
perhaps require that they all succeeded, but that might be excessive).

The delay argument just doesn't fly with me, honestly.  We still only
push out to mirrors once a day anyway, no?  Not to mention that it's
unstable, yadda, yadda.


Attachment: signature.asc
Description: Digital signature

Reply to: