I'd also like to clarify this issue
in case anyone else has made similar "assumptions" that have
been in place for many releases.
We in WTP have many extensions to the
base emf Resource class, and have several mechanisms that depend on the
logic and timing of Resource.isLoaded() including our synchronized loading
adapter that serializes multi-threaded resource loading.
We were hit by the assumption "Resource.isLoaded()"
will come back as false if the file uri we are passing doesn't exist on
disk. - This will now always come back as true.
Just wanted to keep everyone informed on some late breaking developments
It was discovered that a change in EMF exposed an issue in WTP, where we
were making certain assumptions that we should not have been (according
to the EMF team :)
The EMF team has kindly agreed to back out that change and produce an RC1a,
which would require no change from WTP.
This EMF RC1a should be ready late this evening or early in morning (depending
on your time zone). And, BTW, the change in EMF should not require anyone
to have to recompile with RC1a.
This will allow us to produce a Ganymede RC1 (with EMF RC1a) in the safest
possible way, and give us (WTP) time in the days ahead to adjust to the
EMF change in RC2.