Magnus Hagander wrote:
> Also, AFAIK we don't *have* a "message type" at this point (one of
> the things said mythical project wanted to look at), so the only
> thing we could really filter on would be the whole text of the
> message, no?
We have SQLSTATE, but this seems to be one of those situations where
we've been sloppy about using the right value. We seem to be using
'08P01' (protocol_violation), which is also used for finding the
wrong bytes on a working connection. It seems to me a broken
connection is exactly the case where you would expect to see '08006'
(connection_failure). FWIW, there are also specific exceptions for
rejecting a connection attempt, and for attempting to send something
when no connection exists.
We don't need to invent new mechanisms for categorizing messages; we
just need to start consistently using the one we have correctly.
-Kevin