Andreas Pflug wrote:
> Tom Lane wrote:
>>
>> Almost everything I just said is already how it works today; the
>> difference is that today you do not have the option to drop t1 without
>> dropping the sequence, because there's no (non-hack) way to remove the
>> dependency.
>>
> As far as I understand your proposal I like it, but I'd like to insure
> that the situation where a sequence is used by multiple tables is
> handled correctly. There _are_ databases that reuse a sequence for
> multiple serial-like columns, and pgadmin supports this (including a
> pg_depend insert, which would need a version dependent fix).
>
If we were implementing serial from scratch, I would be arguing that the
underlying sequence should be merely an implementation detail that should
be totally hidden, and sequences used explicitly should be kept as a
separate concept. Then many of these problems simply wouldn't exist. I
realise that might be difficult to get to now :-(
cheers
andrew