Re: Running gump

[Mark Wielaard <mark at klomp dot org>]

Hi,

On Fri, 2005-01-28 at 06:00 -0800, Anthony Green wrote:
> gump <http://gump.apache.org/index.html> is a tool for building free
> java software for purposes of integration testing.  It currently tries
> to download and build 812 packages.
> 
> Last night I wrote a simple cron job script (attached) that:
> 
> 1. builds FSF HEAD gcj
> 2. builds ecj from eclipse.org cvs
> 3. builds java-gcj-compat from cvs
> 4. runs gump with the above
> 
> Only 60 packages built on the first go.  I think there are some very
> simple things preventing others to build.

Nice!
You might want to take a look at the configuration of the Apache/Kaffe
setup of gump: http://brutus.apache.org/gump/kaffe/
That is currently at 26%. You will need some patches that have already
gone into GNU Classpath and Kaffe, but haven't made it into libgcj yet.
(There seems to be some reluctance to add anything new to libgcj now
that the verdict is that we aren't release critical anyway.)

> Unfortunately, I don't seem to have an option to conveniently host
> browsable logs, since they are so big.    I've temporarily placed a
> compressed archive here for the curious:
> http://people.redhat.com/green/log.tar.bz2
> 
> I'll figure out how to host this properly (maybe gcc.gnu.org?).  And my
> cron job is already set to go daily.   Any suggestions on tweaking the
> script are welcome.

We could probably host it on developer.classpath.org. That will soon
move to a bigger and faster machine anyway. But Gump still needs a
couple of non-free binary blobs to function correctly (the blue things
in your log) and that should be really fixed first.

Cheers,

Mark

Attachment: signature.asc
Description: This is a digitally signed message part