RE: Dynamic sized multiple occurrence data structure

Believe it or not, this particular app uses 25 files to build the result
set that is sent back. Not all files are used for all request types so
each request does not open all 25 files. I found it easier (from the
perspective of someone who can use SQL put I would not say I am an
expert) to code this in "standard" SETLL/READs and CHAINS then some hugh
complex SELECT statements. I have seen some hugh SELECT statements that
work but 6 months later no one can understand. I have done a few where I
build the result set into an externally defined QTEMP file then used a
SELECT on the QTEMP file with the ORDER BY clause. In this case we were
reusing code that already existed for a green-screen app that put the
data into an array and sorted it to them be put to a subfile. Just
changed the "put to the subfile" into a "put to a MODS" for the stored
procedure

Jon.Paris@xxxxxxxxxxxxxx 12/15/2006 5:45 PM >>>

And in this particular app I actually have an array and a MODS. I

load
the array because I then sort the data before loading it into the
MODS.

Why not use qsort to sort the MODS directly?  Or alternatively use a
pointer
to base the MODS on the storage used by the array?  That approach would
help
both in terms of the memory footprint and speed since there's no need
to
copy the data.

But why is an ORDER BY not being used in the SQL to avoid the need for
sorting - or am I missing something? 

Jon Paris
Partner400

www.Partner400.com