Hi Denis.

>What I'm wondering about it this. Many indexes were defined depending on some processing that has to be done in the application. Would'nt it be worth it to create a data engine class? Doing that could improve the overall performance of SFR by eliminating some relations that are not necessary for the reports.

I don't think that would improve performance; relationships defined in REPMETA don't really slow things down. Even in a complicated data setup like ACCPAC, with over 300 tables and thousand of relationships, it takes SFQuery only a second or two to generate the SQL SELECT statement for a report.

>If I use another data engine class do I need to re-create RepMeta?

No.

>To answer your question when I look in RepMeta there are relations that were created. I assume that those are correct. So I also assume that the relation collection is also set up properly.

Then the quickest solution is to bump up MVCOUNT.

Doug