>>>>>>>>>Why CursorFill() is faster than CursorRefresh() in a CA?
>>>>>>>>>Note that both functions returns the same number of records and uses the same SelectCmd...
>>>>>>>>
>>>>>>>>CursorFill()
>>>>>>>>create the CA.Alias dbf
>>>>>>>>
>>>>>>>>CursorRefresh()
>>>>>>>>1. create a temp dbf
>>>>>>>>2. copy temp.dbf into the CA.Alias dbf
>>>>>>>>
>>>>>>>>the step 2. can require enough time,
>>>>>>>>and this force the dbf structure to not change
>>>>>>>
>>>>>>>Hi Fabio,
>>>>>>>
>>>>>>>are your shure?
>>>>>>>
>>>>>>>the DBF() is someting like SYS(2015)+'.DBF'
>>>>>>>I can see DBF(CA.ALIAS) change with every CursorRefresh.
>>>>>>>I run in this indexing a CA / View after Requery / CursorRefresh
>>>>>>>
>>>>>>>Agnes
>>>>>>
>>>>>>try to attach the CA.alias to a Grid.
>>>>>
>>>>>Done
>>>>>DBF() changes with every cursorrefresh Simple to see on Statusbar
>>>>>
>>>>dbf here does not change.
>>>>
>>>>>No idea why the grid should make a difference
>>>>
>>>>Grid do not like if recordsource file handle change
>>>
>>>I have no idea what you are refreing too.
>>>I use a CA bound to a grid with Cursorrefresh since the time the CA was introduced. I have replaced the views (formerly used) with CA and have no problems with grids at all.
>>
>>That is what Fabio means - as you have no problems with grids, CA must at least "shimmy" the "Safe Select" needed for normal cursors - you have to zap the Bound source and append from the new data...
>
>So this means the grid uses a copy of the DBF() ?

To have a good performance, the Grid internally maintains a list of records displayed
and other structures related to the grid's recordsource dbf ;
the grid uses the file handle opened by VFP.

If the file disappears from the file system, the RecordSource becomes invalid and VFP,
instead of recovery the situation, VFP takes away from the grid.

Try to ZAP,PACK,CursorFill() a Recordsource ...

Then a cursor binded to a grid must maintain the appropriate file dbf