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

Re: Let's remove mips, mipsel, s390, ...

On Tue, Feb 22, 2005 at 11:22:37PM +1100, Hamish Moffatt wrote:

> > "Can" and "should" are different stories. 
> > When there's a missing build-dep on one arch, it might make sense to stop
> > that package from being distributed for other archs, so they don't waste
> > their time on that. 
> > You can't do that on a per buildd basis. 
> True. But even what I proposed is a big improvement on the current
> situation.

As Wouter wrote in another mail: sbuild has its own implementation for that.
Maybe you want to file a bug against sbuild then for a better

> > Same for supplying additional build-deps that have already been solved on
> > faster archs. 
> Do we still do this? Aren't missing deps just sent back as FTBFS bugs?

Erm, sort of... of course FTBFS are filed. But when a (new) build-dep is
missing and this is discovered on one arch, it is a stupid thing to let it
fail (and waste time) on all other archs as well, just to send in multiple
FTBFS bugreports for all archs. IMHO, it would be smarter to just file a
single FTBFS bug and add the missing build-dep to the other archs. Sure, it
is arguable if that packages should be build anyway, but it would save time
on the buildds. 

> > Have you ever watched several buildds doing their work to give qualified
> > comments on that issues? If so, join the group at #multibuild... 
> No, but I don't see why I shouldn't comment anyway.

As I said: if you have valuable comments/contributions, those people there
might have an open ear to you. 
It is known (at least to some persons) that the current buildd system has
drawbacks, but that's not a reason to drop some archs from the release.
Sadly, many DDs don't have any/much clue about how buildds actually work.
Good thing is, that the number of DDs that have, has increased over the last
few years. But still, there are many "discussions" floating around that are
based on non-knowledge about buildds.

Ciao...              // 
      Ingo         \X/

Reply to: