Jim,

>I agree with this point. I call it the Kavorkian pattern, never destroy an object, ask it to destroy itself.

>My point on the RELEASE command is simple, If I say RELEASE MyVar I expect MyVar to be released. I don't expect any other variable to be affected by the release command. Having the release of a memvar cause the release on an object when there are other vars referenceing that object just makes no sense to me. It strikes me as a major violation of encapsulation and an introduction of very tight external coupling.

Jim, I suspect that you're not alone in this matter. Though you might have a point here, It would be nice if it would be possible to force an attached object to release, maybe with an extra clause (e.g.: [FORCE])

But as Jim stated (means: I haven't checked myself) MS has promoted to use the RELEASE command instead of the method. I think this statement is a serious error that should be corrected in upcomming documentation (like a helpfile or developersguide).

Another change that definitely has to be applied is a proper description in the help file. I have to agree with Jim N that there many pitfalls for a newbie because the helpfile shows serious shortcommings. Especially on this release matter there is no clear explanation how it actually behaves when it comes to objects.