I vote for generic queries in the tcl files.
I think Alfred's point is a good one,
there is value to being able to write
static code analyzers but I don't think
sql in the tcl files is a significant roadblock
to doing so (in particular, you need to have a pretty
smart parser already to deduce the fully qualified
qeury name and pulling the sql out at the same time
is pretty trivial relative to that problem).
Given that to date we have not done anything fancy with
reusable queries and that there is no support for
paramaterizing queries (and I agree with lars it's better
to put the query in a tcl proc and have that be the reuse
mechanism), I just don't see a material downside to
doing this.
