>I agree with you Dragan. I'm happy to have the debugger too, and I gladly put up with it's minor glitches - I can't imagine life without it.
>
>I do have a theory that the problem only happens when using multiple data sessions. Just a theory.

I may only add to your theory. I think that if ever we try to release a datasession which is in less-than-perfect state (not updated buffers, a primary key dangling somewhere), the automatic release, while doing the "forget it all" operation right, does not complete the cleanup of memory, at least not good enough. It leaves a landmine somewhere, and we're actually lucky that it still works for quite a while after that. Maybe it takes a dozen of these landmines to get a real good C005 :)

I know, that's the stick'n'rope explanation, but the best I can come up with. Reminds me of Atari ST's memory map, where one region was marked with "here be dragons".