Hi Felix,
comments inline
Felix Knecht wrote:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hi all
Following the 'Studio breakage fixed?' thread we should decide about how to use
the CI builds for. ATM the projects are
build (every 30 minutes) but only if there are svn changes made to the specific project.
I think this is good.
Doing so there can be a big
delay before noticing the a e.g. change in shared have impacts on
apacheds/studio/... It will be noticed next time
apacheds/studio/... is built. This will in most cases happen when a commit is
done to one of those projects and not earlier.
We can change this by forcing CI to build the project each time regardless of
the latest build state (ok, failed, ...).
Well, the recent problem we had (studio breakage due to shared being
modified) were pretty much due to some laziness and misunderstanding of
my part. First, I forgot to check that the part IO modified in shared
impacted or not studio. This is for the laziness part. Then, yesturday,
after having done some more modification, I did check that studio was
running fine, and it did... until the next 30 minute CI build :). The
reason is that the way Studio is build was badly understood by me : I
thought that studio was using the local snapshots first, and then look
up on the remote repo. Bad bad bad : this is not the case. So when I did
my first build, it took the oxylos shared jars, and all went ok. But as
I modified shared, those snapshots were modified, and broke the build.
Ok, now, what can we do to avoid such problems :
1) being less lazzy. Working on it ;)
2) the local build should check if there are not more recent local
snapshot before switching to the repo's snapshot. Doing so will allow us
to check that the modifications are ok, and don't break the build.
Doing so I'd suggest changing the build schedule per project at maximum twice a
day. Otherwise will run probably into a
queueing problem because projects are still in build process while already
beeing queued again.
I would rather keep the current system. it's working well, and assuming
you are not committing and running to bed, you have 30 minutes to fix
the potential breakage.
Does it make sense ?
--
--
cordialement, regards,
Emmanuel Lécharny
www.iktek.com
directory.apache.org
