Hi Hilmar
I was surprised by the 1.25 seconds but that was the results I got consistently when checking with the coverage profiler. So I simply removed that one line from the code and the speed went up dramatically.
I would agree that doing the flush after a batch would be good.
Simon
>>Hi Hilmar
>>
>>I gave up using Flush because it was taking 1.25 seconds per record in VFP 7.0 on a Dell PIII-450mhz Win2k 100meg Ethernet network. So if I updated 100 records it took over two minutes. However, removing the Flush the update was done almost instantly.
>>
>>I understand the risks but the users were complaining about the slow speed of the application. So I would only use the flush command in some very critical sections.
>
>I didn't check the time for an individual record, although meseems it should be considerably less than 1.25 seconds.
>
>However, when a batch process changes several records, I would recommend (as a compromise between speed and safety) to do a single FLUSH at the end of the batch process. Of course, just closing the files after the batch update does the same.
>
>Hilmar.
Simon White
dCipher Computing