On 7 Dec 2001, mcornell wrote:
> We're new to postgres and using it on Linux and OS X. We've had to
> change our code to work around what looks like a bug: When a table
> that contains a SERIAL column is dropped, the corresponding sequence
> for that column is *not* dropped. This is a pain, because we are
> trying to support multiple RDBMSs, and only postgres requires this. It
> seems like it should be keeping track of which columns are serial and
> drop them when the table is dropped. Please tell us: Any plans to
> address this in the near future? Thanks!
There's been discussion about this in the past. This will probably be
dealt with as soon as we start doing a more meaningful list of what
objects depend on what other objects. Right now, there's no good way
to make sure that the sequence isn't being refered to from something
else, and I'm not sure what the correct behavior would be in that
case (probably depends on the whole restrict/cascade thing I guess).