> OK. Keep in mind it is possible the error was completely
> bogus. I had been testing the Import Existing Project
> feature and I imported into my Runtime Workspace the
> Subclipse UI project. The same project that was running in
> the runtime workspace. So when that process rebuilt all of
> the classes, it messed up the runtime itself. Since cleaning
> everything up I have not had this problem.
>
Sure, I would prefer to believe it was a bogus ;-)
However, it also might be a early warning of a problem which really exists
there,
it just does not occur in small scales of our testing scenarios.
These nasty multithreaded things ...
I'm not that deep into eclipse internals, but according to that stacktrace I
deducted that there might be a real problem there.
When the pending cache is being flushed, the setSycnInfo is being called
which starts an operation which acquires a workspace "lock" on some
resource.
Seems the locks could be nested when they have common root (project).
What probably happened in your case was that the setSyncInfo was trying to
lock workspace on "SubversionBook" but some other thread already
Acquired lock on project "SubversionBookFeature".
I'm not sure whether this really happened, it's just my deduction.
I'm also not sure how to solve it if it is that problem.
Do you guys have some experience with these eclipse "operations" and/or
synchronizing eclipse threads ?
Or at least some hint where should I take a look ?
Martin
Received on Wed Jul 13 02:09:45 2005