another classgraph bump

[cdietrich]

also using 0.0.0 as @merks keeps doing updates too often
(our 2.36 release is broken, but i wont have time for a hotfix) :(

[merks]

Sorry, I didn't consider that this might break something. I could have tried to back out the changes but now Orbit is also released. Sorry sorry.

[github-actions]

Test Results

  6 460 files  + 4    6 460 suites  +4   3h 10m 51s ⏱️ + 9m 48s
 43 240 tests ± 0   42 657 ✅ + 1    583 💤 ±0   0 ❌  -  1 
170 103 runs  +10  167 736 ✅ +21  2 354 💤 +3  13 ❌  - 13 

Results for commit 8b7aaa3. ± Comparison against base commit 0db6956.

[LorenzoBettini]

also using 0.0.0 as @merks keeps doing updates too often (our 2.36 release is broken, but i wont have time for a hotfix) :(

What do you mean by broken? It cannot be installed?

[cdietrich]

no, the new xtext project wizard will produce a non working target platform

[mehmet-karaman]

no, the new xtext project wizard will produce a non working target platform

Maybe there is a previous version of the orbit which we could refer?

[cdietrich]

For the release it does not matter
As the orbit url and the version is hard coded

[merks]

For future reference, the Platform's RC2 build is at offset -1 which means their contribution is Friday the week before everyone else contributes. They do the build for that on Wednesday. The Orbit, EMF, and ECF milestones are due Tuesday/today. Given this is RC2/final, Orbit, EMF, ECF must (should) provide final content, i.e., release content. To avoid last minute stress, the Orbit release build was done yesterday. I could have backed out the version if I were informed sooner. The update was done by this commit one week ago:

eclipse-orbit/orbit-simrel@036cf7c

In general updates happen the day after something becomes available and updates generally continue until the Platform's RC2 week.

[cdietrich]

this basically destroys the relase flow we did the last 8 years.
we alwayst built releases at RC1

and that was at a time when i still had day time to work on xtext

[merks]

Unfortunately I'm not aware of every flow of every project. It looks like the release was produced 2024-08-25 so if I'd produced Orbit milestones more rapidly toward the end of the release cycle, the Orbit change from 2024-08-21 would have been consumed by your build. Live and learn.

One could subscribe to the repo where the changes happen:

https://github.com/eclipse-orbit/orbit-simrel

One might do less hard coding, though I don't know if that's feasible.

One might time releases around Orbit releases if your release content depends on exact versions:

This pattern has been in place for quite some time. I suppose the important change is that Obit updates are semi-automatic such that there typically changes right up until RC2, unless someone makes me aware of a change that is problematic so late in the cycle.

[cdietrich]

i guess we just need to live with stuff exploding.
or have other people than me building....

[LorenzoBettini]

@cdietrich we're still on time to release a fix, aren't we @merks ?
By the way, Maven artefacts are not affected, only p2 stuff: we could even just release another p2 site (the build qualifier will be increased) and skip the Maven deployment (though our Jenkinsfile should be updated in that respect). Otherwise, we fully release 2.36.1.

In general, I had already proposed not to use fixed versions of such dependencies in the target platform ;)

[cdietrich]

I don’t think we did a dot release for years
And I can’t contribute anything

[merks]

The RC2 contributions are not due until next Wednesday so from that point of view, no deadline has passed.

I am hopeful that at some point PDE will support version ranges in the target platform. Perhaps that will be useful to avoid locking in one specific version while still specifying some version constraints...

---

No to downplay the problem, but is this only a problem for people who create a new project with the wizard and then it's a problem easily addressed by correcting the hard-coded version number? PDE even has an automatic fix for that...

[cdietrich]

yes this is cause my solution would be "live with it"