>>>>
>>I think you meant Gregory.
>>
>>what you're doing (by design, I guess). So the grid will obey some internal clock and VFP will decide it's time to refresh. It will visit each visible records, read the values and refresh one row at a time. At some point this will clash with whatever you're doing in grid's alias.
>
>Nothing to do with the clock - I guess those things are event driven - asynchronously

In the case of grid, the event may be some system timer firing. That's probably required to stay compatible with the old browse command, which had some refresh dependent on set reprocess or some such. Seems to ring a little bell somewhere, that these may be connected. And since a browse is a grid, and there are no timing properties exposed, I think the grid refresh is independent from the form's refresh (i.e. it may fire on its own and it may not even count as a refresh). The easy way to check this is to have some code called from a dynamic* property, and have that code log the stack and timestamp (in milliseconds or finer). That would give us some clue as to when does the grid really refresh.

I know I'm avoiding any operation on grid's alias while the grid has the focus; I either use it again under a different alias (if I need to write back to it) or I do some SQL, which is independent from the record pointer in the current alias.