Thread: NOTICE vs WARNING resolution

NOTICE vs WARNING resolution

From
Peter Eisentraut
Date:
Following up to the discussion a few weeks ago and in accordance with the
criteria developed there about how a message should be classified NOTICE
or WARNING, I have identified the following cases that ought to be
reclassified:

change WARNING to NOTICE:

table "%s" has no indexes  [during REINDEX]

change NOTICE to WARNING:

changing return type of function %s from "opaque" to xxx
INTERVAL(%d) precision reduced to maximum allowed, %d

I also thought that

copyObject() failed to produce an equal parse tree

should be made an elog() because it's an internal error.


Objections?

-- 
Peter Eisentraut   peter_e@gmx.net