Comparator errors in I20241106-1800

[jarthana]

These are expected changes due to the compiler changes via eclipse-jdt/eclipse.jdt.core#3254

[jarthana]

Copying @srikanth-sankaran

[srikanth-sankaran]

I will confirm shortly. At first blush nothing looks amiss. Compiler has stopped using secret variables for yielding the value of a switch expression and directly leaves it on the operandi stack. That accounts for stack size changed and variable index changes in the byte code.
Ri
I'll confirm once analysis finishes.

Who can follow up with a bundle update ? I don't have the commit rights.

[jarthana]

Not sure @iloveeclipse is back yet. I guess either @akurtakov or @HannesWell can help.

[akurtakov]

Just say here that it's as expected and should go in and whoever is around will push it.

[github-actions]

Test Results

   285 files  +  6     285 suites  +6   49m 9s ⏱️ +40s
 3 586 tests ±  0   3 510 ✅ +  2   76 💤  -  1  0 ❌  - 1 
10 950 runs  +194  10 719 ✅ +167  231 💤 +28  0 ❌  - 1 

Results for commit 40969fb. ± Comparison against base commit f3ba023.

[srikanth-sankaran]

Just say here that it's as expected and should go in and whoever is around will push it.

I have studied these :

1.  eclipse.pde/apitools/org.eclipse.pde.api.tools/.polyglot.META-INF
   no-classifier: different
      org/eclipse/pde/api/tools/internal/builder/IncrementalApiBuilder$ResourceDeltaVisitor.class: different
      org/eclipse/pde/api/tools/internal/util/Signatures.class: different
    The main artifact has been replaced with the baseline version.
    The following attached artifacts have been replaced with the baseline version: [sources]

No harm, no foul.

As alluded to above, the unnecessary store into and load from secret location used for yielded values in a switch expression go away and this changes byte code offsets and sometimes bytecode size/shape too.

So we can proceed with the version bump.

Thanks for your help with this, gentlemen.

[srikanth-sankaran]

Test Results

   285 files  +  6     285 suites  +6   49m 9s ⏱️ +40s  3 586 tests ±  0   3 510 ✅ +  2   76 💤  -  1  0 ❌  - 1  10 950 runs  +194  10 719 ✅ +167  231 💤 +28  0 ❌  - 1 

Results for commit 40969fb. ± Comparison against base commit f3ba023.

I have no idea how to interpret this, but overall https://download.eclipse.org/eclipse/downloads/drops4/I20241106-1800/testResults.php looks good with pde entirely clean.

[akurtakov]

New I-build started https://ci.eclipse.org/releng/job/Builds/job/I-build-4.34/122/

[laeubi]

So we can proceed with the version bump.

We need a new compiler deployment as otherwise now PDE verification builds fail.

[srikanth-sankaran]

So we can proceed with the version bump.

We need a new compiler deployment as otherwise now PDE verification builds fail.

Sorry to ask a dumb question: what exactly if anything needs to happen from JDT Core side for this ??

[HannesWell]

So we can proceed with the version bump.

We need a new compiler deployment as otherwise now PDE verification builds fail.

Sorry to ask a dumb question: what exactly if anything needs to happen from JDT Core side for this ??

I assume you have to run the JDT Jenkins job copyAndDeployJDTCompiler:
https://ci.eclipse.org/jdt/job/copyAndDeployJDTCompiler/

Probably with the parameters equals to
https://ci.eclipse.org/jdt/job/copyAndDeployJDTCompiler/177/parameters/
except that buildid should probably be I20241110-0600 and ecjversion should be 3.40.0.v20241110-0909.

The versions are the latest ones, maybe a bit earlier would also fit, but it should be at least the version that contains the change causing this.
In #1467 I'm trying out the latest snapshot.

[srikanth-sankaran]

@jarthana - FYI - Are you able to help with this ? TIA

[HannesWell]

@jarthana or @srikanth-sankaran any updates on a new deployment?

[jarthana]

Sorry, missed this one. I just tried with build I20241111-1800 but looks like all nodes are down. Let me follow up that.

[jarthana]

The nodes have been brought up again and the maintenance work on repos is over. Now, the job is finally complete (I used build I20241111-1800, though).