In my experience, CLEAR ALL does a pretty good job of cleaning up memory (particularly as it relates to memory handles).

Of course, you have to do it in a part of your program where you don't destroy your application's infrastructure ... Usually within the main program loop and where you can serialize and (re)initialize your app's globals. Requires some upfront design work.

Sort of the alternative to VFP garbage collection (which doesn't appear to exist).

You should, of course, do a small test in your environment to see if there is any impact on RAM; SYS(1016) and SYS(1011), which CLEAR ALL will reduce, refers mainly to VFP memory management.

>Hi all,
>
>Got a bit of an issue running some of our applications on Terminal Server.
>The kit is a multi processor Xeon with 8Gb ram, and can have 25+ users connected running a VFP app sharing data currently on the same machine.
>
>All reg hacks have been made etc, and the IT people who are supposed to know about these things have checked the TS configuration.
>
>However, when some of the apps run large queries/views the amount of Ram used by the application shoots up from approx 30Mb to 150Mb. Fair enough, perhaps the cacheing/buffering kicks in. Unfortunately this does not appear to get released wvwn when the cursor is recreated with a much smaller dataset.
>
>Once this happens to a large number of users, performance degrades exponentially, as if the OS was swapping out to the hard drive.
>Given the amount of Ram, I dont see why this is happening, and I have no idea how to recover that Ram from within VFP. Before anyone says anything, Yes we do use Sys(1104) extensively.
>
>Someone has also informed we that VFP cannot address the memory beyond 4Gb under these circumstances, though I can find nothing on the web to support this claim.
>
>Regards
>
>Malc