Bob,

I was too lazy to go back and read the whole thing, and I certainly agree that Rick has the best command of this topic that I have seen here.

But your final line questioning if a VFP 3-tier solution *can* be implemented is answered at this time with a resounding NO, I believe.

I *think* Rick had (and can still, of course) done so BUT it invloved other components as well.

My understanding of about the best you could achieve with VFP and multi-tier is a tier-1 of VFP, a tier-2 of VFP in-process (DLL) server (that is, VFP 2nd tier residing on same system as client, and a non-VFP thrid tier. The non-VFP thrid tier would be the DB component and something supported by MTS (or whatever did the trick).

Good luck,

Jim N

>>should conform to Single Threaded Apartment model threading. VFP6 actually
>>does conform to this model, but has internal blocking to prevent simultaneous
>>operation of method/property access.
>
>WOw... glad to know someone understands this.
>
>>VB and VFP only simulate this by loading mulitple areas in memory with
>>the same code and their own private Thread Local Storage.
>
>I see...
>
>>From what I can see the blocking is required to protect those globals from
>>getting hosed by simultaneously running instances. Globals being things related
>
>
>So, if they can block this an allow sumultaneous 'psuedothread' is MTS still not needed?
>
>As you said, it seems to be the MTS is really only needed when doing homogeneous transactions or complex multi-object transactions?
>
>However, isn't another advantage to MTS the role based security?
>
>Hmm... what to do... we are told three tier... but still deciding on how the middle will work. Is it possible to create a scalable 3-tier in VFP without using MTS or any other middleware?
>
>BOb