>>Hi Gregory,
>>
>>>&& error 107: Operator/operand type mismatch
>>>&& both n and Offset are initialized and numeric
>>
>>What I've seen on Windows 98 machines is that a numeric value suddenly came up as a currency value which caused a similar error.
>
>
>Win2K or XP
>
>
>>>local i
>>
>>"i" isn't a particular stable choice for a variable name. One reason is that VFP reserves the letter A-J as aliases for the first ten work areas. If you are using INTL "i" also the name of the function that performs the translation.
>
>I'll have a look at that (am not using INTL) but I doubt that - it would confuse an alias and a variable
>
>
>>> local cs && here error 1960: illegal redefinition of variable cs
>>
>>Can you create a LIST MEMORY dump when this error comes up. Seems like CS has been used before thereby forcing VFP to create a private variable.
>
>I have not been able to reproduce it in IDE. But I'll add a List Memory to the error routine : GOOD idea
>
>My latest actions have been - and apparently the frequence has diminished
>
>(1) release arrays of objects either by a release statement or by setting array[m.xx] = null
>Reason: when not releasing arrays that contain objects - trusting the endfunc - I have noticed that the objects only get destroyed when there's an idle state. (the inits are cpu intensive)
>
>(2) setting autoyield to FALSE at the beginning of the init() and resetting it at the end
>
>Thanks,

I think you may update this thread with your resolution of the problem.