Revert "Wait for deferred PDE classpath initializer before initial reconcile"#2392
Conversation
reconcile" This reverts commit 1f01b41. The code in PDE that used ClasspathContainerInitializer.class family is gone, so this code is meaningless now and only increases code complexity. See eclipse-pde/eclipse.pde#1898
|
I've seen that the PDE doesn't use theClasspathContainerInitializer` family anymore, so I've removed it in JDT UI, but I'm not sure if we need yet another fix to address the problem. @merks : if you have time, could you please check build https://download.eclipse.org/eclipse/downloads/drops4/I20250729-1800/ (which does not contain this PR), if you can observe the problem of #2069 ? I can't, however I wasn't able to reliably reproduce it before either. |
|
Yes, as you've probably noticed via the PDE PRs, @laeubi has been reworking the implementation details such that the classpath information is cached and is readily and cheaply available on startup. That of course makes the waiting a no-op. This morning I did extensive testing (many dozens of restarts) that the IDE restarts well with some open Java editors and open .classfile editors all with breakpoints in them. If your change was in I20250729-1800, I've already tested that and saw no problems with phantom errors so all appears to be on a forward progress track! |
|
Thanks for the heads up BTW! |
This reverts commit 1f01b41.
The code in PDE that used ClasspathContainerInitializer.class family is gone, so this code is meaningless now and only increases code complexity.
See eclipse-pde/eclipse.pde#1898