Test process

Process

We center Mer contribution around our Gerrit instance and git, http://monster.tspre.org:8080 . Every single package is in git and the Core (+ ports) is described in xml files inside a git repository. It is possible to monitor gerrit events with 'gerrit stream-events'

Terms

QA build server: OBS that does builds to verify if sources work okay

Code review: ssh -p 29418 username@monster.tspre.org gerrit review PATCHSET -m "Message" --code-review SCORE

When a change to a package is proposed

Step V: Validate packaging

. Validate proposed changes + packaging as being OK

Step X: Checking if it builds on it's own

1. Create unique OBS repository, REPO

2. Create package (basename of project/packagename) in REPO, by linkpacing fakeobs:Core:ARCH packagename

3. Locally, check out git (GERRIT/project/packagename) and merge proposed change from branch. Change ref can be found with 'ssh -p 29418 username@monster.tspre.org gerrit query status:merged --patch-sets --format JSON' (status:merged can be any kind of query) and it will give you patchsets + refs.

4. Upload sources to OBS package in REPO

5. Point to repositories in REPO:

for each ARCH, ARCHSCHEDULER, add in

<repository name="Core_ARCH">
 <path repository="Core_ARCH" project="fakeobs:Core:ARCH" />
 <arch>ARCHSCHEDULER</arch>
</repository>

6. Wait for build to finish of package for each architecture. For 'succeeded' build, code review with +1 "Looks good to me, but someone else must approve", for failed, code review with -1 "I would prefer that you didn't submit this" along with URL to build log,. For 'excluded', report which architecture in comment with 0, no score.

Home run! Great slugging with that anwesr!