Re: ALTER SEQUENCE

Don Y wrote:
> Bruce Momjian wrote:
> > Don Y wrote:
> >> Bruce Momjian wrote:
> >>> Don Y wrote:
> >>>> Hi,
> >>>>
> >>>> It doesn't appear that there is a way to rename a sequence
> >>>> (ideally with a "cascade" action).
> >>>
> >>> Uh, the ALTER SEQUENCE manual page says:
> >> Uh, the 8.0.3 man page for ALTER SEQUENCE makes no mention of this.
> >> Nor does "\h ALTER SEQUENCE" in psql yield any pointers.
> >>
> >>> Some variants of ALTER TABLE can be used with sequences as
> >>> well; for example, to rename a sequence use ALTER TABLE
> >>> RENAME.
> >>>
> >>> Does that help?
> >> Sure! It appears to allow both the rename and change of ownership.
> >> Thanks!
> >
> > I see the documentation mention added August 1, 2005 byt Tom Lane.
>
> Date tag on the bottom of my man pages is "2005-01-17" -- so that
> explains *that*! :>
>
> In general, how safe is it to use "current" man pages (to
> sidestep these sorts of issues)? Obviously, there will be

Uh, not very safe.

> things in the newer pages that reflect changes NOT present
> in older versions... but, will the documentation updates
> (i.e. this a prime example) outweigh the confusion added
> by documentation for not-yet-existent features/fixes/etc.?)

Probably not. This is an edge case, and often we backpatch changes like
this, though that didn't happen in this case. For example, when I find
doc things to add for 8.2, I add them to 8.1 if appropriate.

> >> Obviously, the documentation doesn't agree with the code :-(
> >>
> >> But, it still leaves unanswered the question of the risk involved
> >> in just changing the name/owner in the system tables...
> >
> > It is best to use ALTER. The only other sure-safe way to do it is to
> > look at the ALTER code and do the same things with the system tables.
> > However, in most cases a system table modification works fine, but I
> > don't recommend it for production servers.
>
> One would *hope* that there was no redundant "information"
> in the tables... but, realistically, that may not be the
> case (efficiency hacks, etc.)

Uh, there is dependency information that might be affected by certain
updates. It isn't really redundant.

> For *this* problem, an obvious solution exists. And, even
> if it didn't, dropping the sequence, recreating it and
> reinitializing it wouldn't be that painful. I'm just
> wondering how aggressive I should be in "tinkering"... :-(

You really have to consult the code to find out.

--
Bruce Momjian http://candle.pha.pa.us
EnterpriseDB http://www.enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +