Hi Hilmar,
>
>Ah, well, I am still working with VFP 6. However, I would like to discuss the advantages of each approach.
Sure, but I am no expert in both areas <g>.
>
>In what version were CursorAdapters introduced - 7 or 8?
8

>Do you have any idea which approach would be faster?
I haven't any real measuring yet. I've done a few measurements on views,
mostly against using tables and/or arrays. As with most constructs, views can be
faster or slower, depending on the circumstances. In most cases I've seen, they were NOT a
benefit, but this was not the fault of the views (CA's would have been wrong there as well)
but of the programmers /"architects".

My guess at the moment is, that the base mechanisms are very similar.
Even setting a CA up reminds in quite a few places to views.

>Although I didn't work with CA myself, I have read the article in utmag. It seems to me that a CA would be more work to set up.
One of my beefs with views was the maintaínance. I was on projects where a lot of views where handled just in code, and this was UGLY. [Perhaps the improved view editor could help there as well]. IMHO a class based solution was missing. I was contemplating to build a bit on the data classes in kilo/megafox, but CA are nowadays probably a better foundation for such work.

>OTOH, it may have the advantage of being more flexible. I assume it is for this flexibility that you would consider CA, right?
Yes, and the reduced maintainance stemming from not having done any "consolidation" work for views. Yup, I've a few lines which reduce the amount of code for views, but it is NOT polished and not even fit to be the common ground for a group of programmers. The building blocks were there (dbc-events for instance), but not time enough or a client with a pressing need to build this. MS giving direction in this area is good in my book.

One thing .net does quite well is distributing the job(s) to be done to a fine-grained set of classes. The cursoradapter being somewhat modeled after .net also gives you a way of slicing problems into similar constructs while still keeping all the benefits of vfp. And switching backends, while possible with views, is easier with CA.

At least with a few demo's, not enough expierience with real app's rolled out yet <g>. Still, you loose quite a bit of the power of each backend by reducing to common functionality.

BTW:
Give vfp9 a work out. You won't need more computing power in your boxes, and [at my own risk] it is used here on some NT4 boxes as well. There are quite a few customers here which still use NT4 and are switching over to XP only gradually. If it is not a specific client holding you at vfp6, you will gain a lot. And the public beta is stable enough to do most of the dayly development if you use SSafe a bit more often. Only danger: you don't want to get back into vfp6, which I sometimes HAVE to do <g>.

rgds

thomas