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

Re: Another pbuilder run finished



On Fri, Dec 27, 2002 at 12:09:40PM +0900, Junichi Uekawa wrote:
> > >   Couldn't be used a more powerful Debian machine? Perhaps this should
> > >   be done with periodicity but, IMHO, a process which takes a month is a
> > >   lot. Reducing it to something like a week or so could be great (I
> > >   don't know really which machine are you using, and what would be
> > >   needed to do this in a week)
> > 
> > We could speed up the runs by distributing them over several
> > voulenteer machines. I have a few gigahertz that I can throw at the
> > task.
> 
> I was thinking about that. If people are willing, that would be nice.
> 
> It is currently running on my personal system.
> 
> The possible drawbacks is that you need root privilage to run pbuilder,
> and running pbuilder over the whole distribution requires a good 
> network access to a mirror, or preferrably a personal 
> full source/i386(or other architecture) mirror.

  About being root, should not a very big problem in volunteers
  machines. At least I can be root in my machine ;)
  And about the network need, that's obvious. But I think that enyone
  interested in this, will also have a good network access.

> 
> I was thinking of some kind of mailing list you can stuff
> random build logs at, so that people can go and check, for a start.

  I think that it could be nice if there is also a little script when
  you can get "packages to built". As a start it should give packages,
  mark packages which build log has been sent, and unmark some of the
  packages, given a timeout, if the log has not been sent.
  Something like distributed.net or seti@home do.

> real failure comes from, however we do need to have a report of 
> some build failure first to notice the problem at all.
> (for example, a broken shared library package or a broken toolchain will
> result in several dozens of build failures, so a build failure bug should
> not always be filed to the package which fails to build.)

  IMO, people interested in this being runned will have the hability to
  check that. At least in a first moment.

-- 
  Jose Carlos Garcia Sogo
     jsogo@debian.org

Attachment: pgpv0MF7mxq5M.pgp
Description: PGP signature


Reply to: