I think the need for this capability has been underestimated. If you are not involved with intensive data processing of large amounts of data then you really do not understand the issues.

We should mention the need to perform processing on large sets of data that reside on multiple platforms. I'm often pulling data from Oracle, DB2 and SQL server. No viable way to deal with that using SP's and forget about using ASP.NET with its current in-memory object model. I have been waiting and hoping that ASP.NET and LINQ will include the ability to span to disk. If that ability is provided then there also needs to be a serious effort placed on improving performance as well.

ASP.NET is my primary development tool and has been since it was first released. To this day I still use VFP for backend data crunching when the requirements exceed .NET’s capabilities. These VFP backend processes run on dedicated servers where the hardware configuration (server/network) have been tailored for the task.


>AGAIN with the automatic spanning of datasets? Good grief, Charlie Brown. Is this truly a "high profile" issue or just your personal Moby Dick? It seems like something an assembly language or C programmer would be concerned about. Not that there's anything wrong with that....
>
>Sheesh, Mike. It's as if somebody keeps saying that Switzerland has the biggest army on earth and if you offer a differing view, somebody responds "AGAIN with the Americans and their guns? Good grief, Charlie Brown". ;-)
>
>As for disk spanning- it's something you took advantage of and used without thought in FP every time you created a cursor or USEd a table.
>
>Some of its uses can (and should) be replaced. For example, "use mygianttable" may not be a resource issue in FP but there are very good reasons to use a paramaterized query (even though a parameterized local query effectively does a "use mygianttable" on your behalf ;-) )
>
>If you're in the habit of using parameterized queries rather than scans or seeks against large tables, conversion to some other platform is easier. In our case, we've used C/S backends and parameterized queries since 1995. Any of that stuff can be ported anywhere.
>
>But there are cases where local datasets carry massive advantage. That thinking may not be popular outside VFP (or inside VFP, it would seem ;-) ) but there are lots of developers out there who will grab hold of the local dataset and run with it as soon as the resource issue is managed.