Re: RPG SQL precompiler

Vernon Hamberg wrote:

4. Re: RPG SQL precompiler (Vernon Hamberg)

I just got bit by this "feature" of the precompiler - have a program
that has embedded SQL - when I put in some SQL CLI stuff, to use a
stored procedure with result set in RPGLE, I got a message saying
that a certain column in a file was an invalid timestamp.

The thing that broke it was when I declared a variable in a
subprocedure that had the same name as a column from a table used in
an embedded select statement - and I also had an f-spec for that
table. What happened is, the precompiler mapped the select field list
to positions in the input buffer - please don't correct me too hard
here if i'm a little off - it's close enough for me to solve the problem - heh.

Vern:

This seems to be a potential spot for a /COPY member. Use /COPY (not /INCLUDE) for the proc statements. The pre-compiler shouldn't see the 2nd variable definition. No?

Or was it the other way around -- you wanted it to see the variable from the proc rather than from the F-spec? Or it needs to see both at the same time -- global and local?

Tom Liotta

When the variable with the duplicate name was present, that field was
completely omitted from the mapping that the precompiler did - and in
this case it moved what it thought was a timestamp to the start of
the field that should have been there - nasty, what?