Hi Mike
>
>I'd change to using TEXT ENDTEXT. You know 15 or 20 macro expansions is slower than 1.

No question about that. But those queries usually run for couple of minutes each, as they are working on datamining insurance data
(even those with a where clause usually involve a lot of data <g>) and so the speed of compiling is not the issue, but filtering out subsets via indexing and rushmore as well as reuse of indices in joins makes my life easier. I should have made more clear that this is not singular data replacement.

Usually a query is written purely in code. Then a similar use case calls for specialising only a fraction (a where clause added only for autocorrelation, siome added fields for company data and such) which are set in specific methods of the classes for that specific use case.

I guess my hesitation on TEXT ENDTEXT at least is based on my begininnings of vfp programming, where it often happened running programs had to be converted into totally different languages - as these usually have no templating system I avoided such use. Slightly irrational case of not using the offered shovel and working with a spoon <g>.

regards

thomas