Erik,
>For me, views cost nothing over tables.

I don't agree that they don't cost anything (speed, losing indexes, needing to index on-the-fly...), but the real point is do you gain anything? Today, not in the event you need to upgrade years from now.

>There are times when parameterized views are essential (forget about the C/S argument) Have you ever tried to build a one-to-many-to-many form incorporating a grandparent-parent-child relationship? If you have and you tried it using filters or linkmaster grids or even SET RELATION, you know that it is a nightmare to the extent of being next to impossible. But using parameterized views makes it simple and fast.

Yes. I had a grandparent-parent-child -- lines for bills for clients -- and (once I figured out linkmaster) found it to be quite elegant. It required almost no coding on my part.

>>...More at http://www.temperedsys.com/UFACS.htm
>
>Not sure I follow this... could you expand? Or should I just follow this link..

Follow the link.

>An intelligently designed UI will provide for this. The first VFP app I built was for an association that had at the time the need for a table with about 50000 records in it. Two years later the size of this table has grown to over 320,000 records. The original picklist design used a grid with incremental search that was based on the table directly. As the months rolled on, this form became too slow almost to the point of being unusable, because of its very design. 320,000 is just too many records to come over the wire, especially when 50+ users are hitting it at the same time. If the form had been originally designed with efficiency in mind, I wouldn't have had to go back in and redesign it. That was the last time I'll make that mistake.

300K records is still small! The problem is in thinking they need to "come over the wire". People are always saying that. At my last gig we had a table w/ 7,000,000 records in it. My PC had a 10Mb card in it. I issued a browse, which came up instantly (under 1 second). I then dragged the bar showing all records and was at the bottom 2 seconds later. My point is, did all 7 million records, all 420 million fields, all 2Gb of data come down the wire? No. I then set up a grid -- with incremental searching using jKey -- and was able to search on all 30 of the indexed fields. How long did the incremental serch take? Before I hit a letter it had already resolved the previous letter. _That_ is an "intelligently designed UI".