autobuild-discuss
[Top][All Lists]
Advanced

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

[Autobuild-discuss] Re: four small patches


From: Ralf Wildenhues
Subject: [Autobuild-discuss] Re: four small patches
Date: Wed, 11 Aug 2010 19:38:09 +0200
User-agent: Mutt/1.5.20 (2010-04-22)

* Simon Josefsson wrote on Wed, Aug 11, 2010 at 01:51:46PM CEST:
> Hi Ralf.  I'm moving this to the address@hidden list, and
> I'm removing all uses of address@hidden address because I
> haven't been using it.

Great, thanks!  You might want to remove the three instances of the old
address which 'git grep bug-autobuild' still produces.

> Ralf Wildenhues writes:
> > attached please fine four small patches that would make Autobuild more
> > useful for Libtool.

> If you want this installed on the build server to get better log output
> for you, let me know.

That would be nice, yes, that way you can take your time rewriting
autobuild.  I intend to start off boldly with something like a couple
dozen test configurations for Libtool in order to have a good base set
of results to compare against, and it would be cool to have decent
reports for them.

Alternatively, it would also be fine with me if the summary pages can
later be regenerated with a fixed autobuild, but without requiring users
to resubmit their logs.  (Chances are, they won't have them any more, or
mess them up in the mean time.  I've done this myself before.)

> > As to tighter integration of Autobuild in Autotools, yes, discussing
> > that is on my TODO list, but I figured getting some experience with the
> > tool first couldn't hurt.
> 
> Sure.  Brian suggested that I extract build timestamp and and build
> hostname from the e-mail logs, and I think that is a good idea.  The
> host triplet is still needed though, but that is less privacy sensitive.
> It is not perfect, as I never intended for e-mail to be the only way to
> deliver logs but it is good enough for now.

Nothing is ever perfect, and I figure that we'll see lots more ways to
improve while using things.

As for the timestamp, the emails might be written an arbitrary time
later than the test, or resubmitted, no?

Cheers,
Ralf



reply via email to

[Prev in Thread] Current Thread [Next in Thread]