"Zeugswetter Andreas SB SD" <ZeugswetterA@spardat.at> writes:
> When those are really truncated ESQL/C needs to set a warning in sqlca.sqlwarn
> though, thus I think the second signature should also have an output flag to tell
> whether truncation actually occurred.
> Maybe this should be kept for a protocol change though, since I would not think
> a NOTICE would be suitable here.
Again, I don't want to invent output arguments for functions today ;-).
I agree that a NOTICE would be overkill, and that we need a protocol
change to implement completion conditions (sqlca.sqlwarn) properly.
When that happens, I think the explicit-cast paths in the coercion
routines can easily call the "set a completion condition" routine for
themselves; I see no reason to pass back the condition one level
before doing so.
regards, tom lane