| From: | Rikard Pavelic <rikard(dot)pavelic(at)zg(dot)htnet(dot)hr> |
|---|---|
| To: | Marti Raudsepp <marti(at)juffo(dot)org> |
| Cc: | pgsql-bugs(at)postgresql(dot)org |
| Subject: | Re: BUG #6489: Alter table with composite type/table |
| Date: | 2012-02-28 21:49:22 |
| Message-ID: | 4F4D4BE2.4040106@zg.htnet.hr |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-bugs |
On 28.2.2012. 15:01, Marti Raudsepp wrote:
> On Sat, Feb 25, 2012 at 15:23, <rikard(dot)pavelic(at)zg(dot)htnet(dot)hr> wrote:
>> This works:
>> alter table t1 add x float not null;
> Peeking at the code, currently any modifications that cause a rewrite
> of the original table are disallowed.
>
> Adding a nullable column without a default is allowed since it can be
> done without a rewrite -- all rows magically get the value NULL.
As shown in example above, I was adding not null column
(but table was empty so query passed)
> I'm not sure whether it's worth complicating code for this. If you
> want to store composite types in tables, I think you're better off
> using CREATE TYPE/ALTER TYPE.
I guess this falls under advanced type usage
(like recursive types - which can be used for lists)
which are not even recognized as important ;(
Regards,
Rikard
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Kevin Grittner | 2012-02-28 22:01:40 | Re: BUG #6497: Error sent to client, but data written anyway |
| Previous Message | Tom Lane | 2012-02-28 21:47:06 | Re: BUG #6496: Why the SQL is not reported as incorrect? Is there a builtin column named "name"? |