Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[orbit-dev] Discussion on coordinates for uploading of approved artifacts

I wanted to open a discussion on this topic as it wasn't resolved in Bug
484633 [1].

When the migration to using orbit-recipes was first starting out there was
a discussion on what would happen in the case of artifacts that couldn't be
found on maven central, or that needed to differ beyond the scope of what 
ebr-maven-plugin could do. It was determined that we would host a maven
repository of approved artifacts at repo.eclipse.org and they could be
consumed from there.

"- What Maven coordinate should we use for published artifacts?
  (libraries may be different from their original, not in Maven Central at
  all, etc.) (eg., coordinates could incorporate a CQ number)"

>From the bug, I agree that if the content is to remain identical then the
GAV should be as well. However, I don't think for the time being we'll be
transferring any artifacts of this case since they can always be used from
maven central.

If the content varies we could make the groupId something like 
'org.eclipse.orbit' and then simply match the artifactId/version/classifier
to what it is in maven central and this could work. Maybe we'd run into
issues with poorly named artifactIds (eg. 'core') so we might consider as
Andreas suggested some combination of 'org.eclipse.orbit' combined with the
original groupId. We could even leave the artifactId name up to the
individual committer and only require the groupId be 'org.eclipse.orbit'.

Thoughts on this ?

The process to upload approved artifacts is nearly in place but I just want 
to be sure on some conventions as once they're deployed (at least for
release repositories), there doesn't seem to be a way of removing them.

Another thing to consider is whether any decisions we make should be
influenced by the fact that we've been asked before about providing a maven
repository of approved artifacts.


Cheers,
-- 
Roland Grunberg

[1] https://bugs.eclipse.org/bugs/show_bug.cgi?id=484633


Back to the top