>>>Hi,
>>>
>>>>
>>>I have done your test and it suggests that my *old* understanding is the correct one and Fabio's understanding is wrong! Agree?
>>>>
>>>
>>>Not really. As I mentioned it doesn't really prove anything. The fact that it doesn't appear to happen in a few thousand iterations doesn't mean that it will *never* happen. And if it *can* happen then any additonal lines between the 'SELECT' and the use of _tally would increase the risk.
>>>
>>
>>Exactly.
>
>How do you mean 'exactly'. What I observed with Viv's code is that the timer's Timer event did not even fire *once* during the loop! It is my hunch that the intended behavior of vfp is that the timer is not actively visited during execution of consequtive inputless commands. You suggest that this intention is not guaranteed everywhere/everytime. Can you write a program that gives support to your hunch?

Hi Peter,

I already lose a sea of time to understand all the bugs that I find.
I could write a program that tries to try that a Timer can also activate him
without DOEVENTS, but if I didn't succeed
it would not be a test some that this can never happen.
( even if the documentation wrote it, it would not be a certainty ! )

If you want to risk and you are convinced that it never happens, not to save _TALLY;
the day that the data of your more important client will be corrupt
you will owe only to curse with yourself.

Fabio