Re: [Subclipse-dev] Subclipse svn operations don't seem to match refactoring operations

> Can you do a similar sequence using plain Java and create the same
> issue? It'd make it easier to look into if we can recreate it using
> the Eclipse JDT refactoring.

Unfortunately, it doesn't seem to repro in Java: in JDT refactors, the
resulting Subclipse operations look like they're happening in the right
order. Now, Java rename refactorings do a different series of operations
than Flash Builder's rename, so I'm not sure we can expect them to behave
similarly anyway.

One difference I can think of is that our refactoring processor creates a
CompositeChange -- that's pretty vanilla, but the contents of the
CompoundChange are all classes we wrote. It's possible that there's
something in ResourceChange we didn't extend, or didn't extend properly,
that Subclipse might rely on. Directly or indirectly.

(I could probably do some debugging in my dev environment, but I've had a
tough time getting it to build, or attaching jars and source to my target
platform so I could set breakpoints and browse. That's made it tough to even
browse the Subclipse code and see if I could make some guesses by code
inspection.)

gse

------------------------------------------------------
http://subclipse.tigris.org/ds/viewMessage.do?dsForumId=1043&dsMessageId=2435863

To unsubscribe from this discussion, e-mail: [dev-unsubscribe_at_subclipse.tigris.org].
Received on 2010-01-08 22:46:56 CET