Re: [glob2-devel] Hudson server and compile errors

[Leo Wandersleb]

Hi Kieran, hi *,

> Re: your home computer failure, I get the same error with USL. Martin
> made some changes recently. Maybe he forgot to compile, or a merge at
> the last second went bad. Either way, you're not alone.

Thanks. Good to know that.

> As for the hudson server error, Steph made some changes for detecting
> the gss compiler. It might use configuration options of a newer scons
> script, so you could try updating it. Here is my scons info that gets
> through to where usl breaks.

I thought so, too but unfortunately I have no root on the hudson box yet. Point
was also to discuss what to do about dependency versions. With a Continous
Integration System you have one reference system that tells you if it is ok to
do a change or not. As we have not agreed on a set of versions of scons, gcc,
... I will not try to make Stéphane "fix" his commit to compile on my hudson but
on the long run I would like to have it this way:

1. We agree on a defined state of the CI-System
2. Commits that don't compile will not be accepted
3. Commits that introduce failing tests will not get accepted

Now hg.globulation2.org is in a broken state for some of us. We need a strict
way to avoid that and DVCS can help in that point. Hudson can track several
forks of glob2 and push to hg.glob2 only those that compile and test fine. What
do you think of this?

Regards,

Leo Wandersleb