>>>I'm in the process of upgrading to a new PC. It's a P6-200 running WIN95b with fat16 (32k cluster size). I have been unable to duplicate the problem I'm having on another new machine with simmilar hardware. FoxPro seems to be fine until I try to use Filer at which point I get 'Insufficient Memory' error. I get the error even when FoxPro is the only thing running. Any suggestions as to what might be causing the problem would be appreciated.
>>
>>You probably need to add a MEMLIMIT command to the config.fpw file. This can have a different effect on different PC's but I have found it is required the more memory you have. It seems without this command, FPW 2.6 will try to grab all of available memory. When this happens, any call to a function that requires additional memory (some API calls require system memory when called, and cannot use the memory that foxpro has assigned itself) will fail because there is none available. MEMLIMIT leaves the system with some memory resources. You might try something like MEMLIMIT = 50, 4096, 8192
>>
>>Bob
>
>Hi Bob, Carla,
> We are having a similar problem with a FPW2.6a application running on Pentium Pro 200's, 32MB, running under WinNT4.0 ( and we are setting MEMLIMIT, which may be questionable under winNT ). The application ran fine on "slower" Pentium machines. We contacted MS and they said the machines were "too fast" for FPW2.6a causing timing problems. We looked at this a little cross-eyed but have no way of disputing it. At some point we are going to test the application on a Pentium Pro running under Win3.1 but haven't had opportunity. MS "suggested" we migrate or rewrite the application in VFP which is the ultimate solution but we will need time for this. Plus, as far as we are concerned, the application ought to run on the machines even if they are "faster" :-). Any other thoughts, information or comments on this matter are welcome.
>
>Bill


I heard about this last year!!! This is really strage, but some engineer at MS suggested putting in delays (like a DO WHILE with a millisecond delay) throughout the application if it detects a faster machine. Sounds bizarre, but it's true. Apparrently, there is some element of parallel processing that is much more (too?) efficient on the faster machines so that certain computations will fail. (I think it's I/O related, but I'm not entirely sure.) It doesn't make a whole lot of sense to me, but that's what MS said.

This was a topic last spring on the foxpro-L mail list. Does anybody else remember this?