>>Hi Steve,
>>
>>First, I know that I wouldn't rate something as "Disagree..." if *I* thought it was not feasible. I'd write a comment to the wisher and I'd withold my rating pending a reply.
>>
>
>I agree with you here (within reason *grin*)
>
>>Now, as to the question of feasibility...
>>
>
>The server receiving the data, and acknowledging that it has successfully stored the data, is one thing. As far as a workstation is concerned, data storage has occured at this point in time. Whether or not it has physically hit the disk is another issue altogether...it is the responsibility of the server; it is beyond our control. It is beyond VFP's control.
>
>I'm sorry, not only do I not see how this is feasible, I'm not convinced it is desirable.

Steve,

Any I/O done by any PC application at any time essentially involves a "contract" with the operating system.
The operating system has accepted the responsibility to either complete the (I/O) request successfully OR let the application know about the failure. It doesn't matter what, where, when, how or why the I/O is done in that the operating system's acceptance of the command constitutes that contract. It is up to the operating system to be able to protect its capability to do this.

I've noticed a recent rash of "Delayed write failure" messages here. This may be a consequence of the OS just accepting the FLUSH and returning (as SUCCESSFUL) to the application! The OS may well have some problem AFTER it returns to the application, but since it has already told the application that the write was OK it informs it using this error message.
And you know, that is tolerable in the case of some batch-style application like Word or Excel - the user CAN save to somewhere else and later investigate and correct the situation. But this is totally unworkable in any typical database system because the user has no alternative place he can store the data (the user doesn';t even know what data is invloved).

As to desireability, of course it is desireable, and not just for FLUSH but for EVERY I/O. Now even the "Delayed write failure" could be corrected to become acceptable. What it would take, and the OS can SURELY do it, is for it to keep the records around until the write has been completed AND to ensure that any additional writes/updates TO THE 'held' records are dealt with in the correct context.

If the OS is going to try to let us have our cake and eat it too, then it has to be sure that the cake can't get mouldy.

cheers again
>
>Cheers to you,