Hi Barbara,

My third attempt to send a note on this will be much more terse, just to save frustration in case of more problems sending. . .

While I feel that the new server "proves" that it is not poor coding at the root of the problem, I do think that the workstations still need to be considered as a potential source of the problems.

1) You own error routine deserves review too. Assuming you have one, it seems possible that an error may be being raised and that your error routine is letting it through. I would certainly examine it to assure myself that this could not possibly be the case. In fact I would make sure that each and every entry to the routine is at least logged so that this can be checked for at any time.
*IF* your routine inadvertently allows some errors without notification (or even advertently does so), then this could be a problem.

2) I, last week, upgraded one of my PCs and found and option in the BIOS ranging from "STOP on ANY error" to "Continue on all errors". If the workstations (or even the server) involved have this option selected (for "continue on all errors" then this *might* be a problem. I have to say might because I have no idea what errors are implicated here, but *IF* they include I/O errors then possibly the BIOS error handler is processing the error (as directed) and therefore not notifying the OS or VFP.

3) The workstations - do all users at some time or other report the problem(s) or is there some subset involved?
If it is a subset, then do the one reporting have different hardware/software of any kind, but especially of video drivers or fancy extended DeskTop features available from some video drivers? Do some have/use Notes or Outlook or IE4 or Win98 beta or things along those lines? Do they all have same level of OS? Of newtork support? Do the trouble terminals all have specific terminal adapter cards (different from OK terminals, or of different release)?

While the network deserves continued suspicion, the ternimals deserve consideration too. I say this because of the commands you have named as skipped, only the USE (and the unopened CDX) would be more or less guaranteed to do actual physical I/O - all the others could well have been serviced from in-storage info where the actual network would not have to do anything.

Good luck, and do let us all know when the solution is found,

Jim N