>>I agree it is a hassle.
>>
>>But under what circumstances would you suggest that it be done automatically? TableUpdate()? Or REPLACE if you don't use buffering?
>
>I'm not talking about buffering at all. This has been the case since many versions of VFP. FLUSH was introduced to allow the cache to be written to disk. I basically don't understand why VFP would create a cache instead of just putting it directly on disk. Some might say it provides a performance issue. But, I'd say that with the product we have now, there shouldn't be any difference if that would be direct.

Precisely without buffering, there might very well be a performance penalty.

Consider this case:

APPEND BLANK
REPLACE Field1 with Value1
REPLACE Field2 with Value2
REPLACE Field3 with Value3
FLUSH

With an implicit FLUSH, each separate REPLACE command would write to disk separately - and that would be much slower than writing to disk thrice, once for each REPLACE command.

Of course, I know that the three REPLACE commands can be combined into one - but do all programmers do this?