At 11:36 AM 2/15/03 +1100, Justin Clift wrote:
>Bruce Momjian wrote:
>>think CASCASE enters into this because of the effect on field1.
>>Comments?
>
>Would it be possible/practical to have PostgreSQL recreate the
>multi-column index, but without the dropped column?
Wouldn't that take a long time in some cases?
I think it's a good idea to throw an error and refuse to drop the column
and index and let the DB admin decide what to do next.
If someone designs a system that regularly drops columns from tables AND
wants indexes on those columns, I'd figure requiring them to drop relevant
indexes first would be a good idea. Of course if they can optionally
configure things (triggers etc) to drop the index when dropping/altering a
column, that would be ok too.
When the admins don't know what they are doing or make a mistake - it'll
fail safe. When the admins know, as long as they are still able to set
things up accordingly, I don't think it's a big problem.
Regards,
Link.