Neither Til nor I want to copy the same (or similar) SQL query
allover the place. As Til notes: OpenACS need to know SQL. They need
to know it better than TCL in fact.
If all SQL queries were moved out into the XQL files we would have
something like a database "persistence layer". Having those queries in
their own respective files would make us look at similarities and
probably reduce the number of different queries and make the toolkit
and the programmer more scalable.
So can the query dispatcher pick a query from a randomly named file or
not?
(Coding SQL is more groovy than coding TCL btw :)
I agree that DML should have a wrapper to support unit testing. But we
don't need a wrapper proc on both TCL and SQL...and there is a unit
testing package for Oracle SQL (by Steven Feuerstein - utlplsql or sth
similar). TCL procs are easier to debug, PL/SQL procs are easier for
migrations. Oh whoever thought that PL/SQL is easier to port to other
databases - I guess he/they were wrong ...