Stephan Szabo <sszabo(at)megazone23(dot)bigpanda(dot)com> writes:
> On Sat, 28 Dec 2002, elein wrote:
>> Almost all of the system generated names, sequences, triggers, etc,
>> have constructed names. $n for constrain names seems like an anomaly.
> I think it's been that way for check constraints for a long time unless I
> remember incorrectly.
I think you remember correctly.
The "$n" convention is somewhat arbitrary, but in my mind it certainly
beats the OID-based convention we have used for RI triggers. For one
thing, if you issue the same table declaration twice, you'll get the
same names associated with unnamed constraints...
regards, tom lane