>>>>That's not the same as saying it's a good as it can be.
>>>>
>>>>Every time any user accesses a form or prg in a VFP EXE, that component is downloaded to the PC from the server. This incurs network traffic for nothing. With it copied to the local PC, there is no network traffic and therefore better performance.
>>>
>>>I'd be very (make that VERY) surprised if that was true.
>>>Have you got some article/paper that confirms that?
>>>
>>
>>I ran a test and posted the results somewhere once. Maybe on the UT. What I am surprised by is how people expect something shared and remote to be faster than something local and exclusive.
>>
>>I mean we all know the data is faster on the local machine, why should it be different for the exe? ;)
>
>While I agree that having the exe locally is the better deployment, I am not sure a prg is copied with each recurring call down the wire: vfp is smart in caching data, are you sure the "data" of the compiled code is never cached from the "database" the app/exe can be thought of ? Have you sniffed or just timed ?
>

I've done both. If it were cached, the timing would only be an initial spike. It remained slow throughout thousands of calls versus the local exe. Filemon shows every exe embedded component is accessed as if it were a seperate file. The consistently slower timing suggests to me that the file is being sent across the wire every time and caching is not affecting it.