Re: Support for detailed description of errors cased by trigger-violations

Andreas Joseph Krogh <andreas(at)visena(dot)com> writes:
> Hi. When working with Oracle it is possible to catch constraint-violations
> caused by triggers using JDBC, but it seems this isn't possible using PG, see
> this thread:
> https://github.com/impossibl/pgjdbc-ng/issues/111#issuecomment-62276464

I'm not exactly following the point. The complaint seems to be that

RAISE EXCEPTION 'ID must be less then 10';

doesn't send anything except the given primary message and a generic
SQLSTATE. Well, duh: it's not supposed to. There are a bunch of options
you can supply in RAISE to populate additional fields of the error report.
For example, you could add

USING SCHEMA = TG_TABLE_SCHEMA, TABLE = TG_TABLE_NAME

if you wanted the report to name the table the trigger is attached to.

So it seems to me this is lazy plpgsql programming, not a missing feature.
It would only be a missing feature if you think plpgsql should try to
auto-populate these fields; but I'd be against that because it would
require too many assumptions about exactly what the function might be
complaining about.

regards, tom lane