Re: buildbot for glibc

[Joseph Myers <joseph at codesourcery dot com>]

On Wed, 11 Feb 2015, Carlos O'Donell wrote:

> That's what we do for Fedora. Our build logs are a combination of the
> glibc logs, plus this snippet:
> 
> for i in `sed -n 's|^.*\*\*\* \[\([^]]*\.out\)\].*$|\1|p' build-*-linux*/check.log`; do

That looks like it predates the move to having proper test summaries, 
instead relying on test failures producing errors from make and using make 
-k check (so would only work now with make -k check 
stop-on-test-failure=y).

I suppose it would be nice to get the web interface to the logs showing 
test failures in some way ("commit X caused test Y to go from PASS to 
FAIL").

> > As with all infrastructure for the project, there is no big supply of paid
> > hacker time to work on this.  It's intended to grow to be of great use to
> > us developers and maintainers, but the only way that will happen is if more
> > folks in the community pitch in to improve and maintain it.
> 
> How can I add more slaves? Can we add more slaves from outside of the google
> network?
> 
> At a minimum I have access to and the ability to add ppc64, ppc64le, s390x,
> s390, and hppa slaves.

Cross-compilation tests that glibc and the tests at least build for 
different configurations would be useful as well - such a test would, for 
example, be able to run the ABI tests and so catch regressions there.  
(The different ABIs at <https://sourceware.org/glibc/wiki/ABIList> provide 
some indication of relevant configurations, although sometimes non-ABI 
differences can make more difference to what sources get built than ABI 
differences.)

I don't know what the machine resources available are - would adding such 
tests (or QEMU-based more complete cross-compilation tests) require 
providing machine resources to run the slaves or just doing the scripting 
to do the builds?

-- 
Joseph S. Myers
joseph@codesourcery.com