Postgres Pro
Downloads
Home   >   mailing lists

Thread: AW: Shouldn't this be an error?

AW: Shouldn't this be an error?

From
Zeugswetter Andreas SB
Date:
> A compromise position would be to allow dropping trailing columns only
> when the column name list is omitted.
> 
> Comments?

Sounds reasonable to me.

Andreas

Re: AW: Shouldn't this be an error?

From
Tom Lane
Date:
Zeugswetter Andreas SB  <ZeugswetterA@wien.spardat.at> writes:
>> A compromise position would be to allow dropping trailing columns only
>> when the column name list is omitted.

> Sounds reasonable to me.

IIRC, there's some interaction between this behavior and the way that
INSERT ... DEFAULT VALUES is implemented; I think DEFAULT VALUES is
parsed as an empty values list and then the trailing-column-omission
code is what actually makes the right things happen.  So changing it
without breaking DEFAULT VALUES is not entirely trivial; it'll take
some changes in the raw-parsetree representation.  There are other
missing SQL features hereabouts also, such as being able to write
a DEFAULTed column explicitly:INSERT ... VALUES('foo', DEFAULT, 42);
and being able to insert multiple explicit rows:INSERT ... VALUES('foo', DEFAULT, 42), VALUES('bar', 99, 44);
I think we should deal with all of these issues at once, which means
it's not something to try to fix for 7.1.  Bruce, would you add a
TODO item?
 * Bring INSERT ... VALUES up to full SQL92 spec.
        regards, tom lane

Re: AW: Shouldn't this be an error?

From
Bruce Momjian
Date:
> missing SQL features hereabouts also, such as being able to write
> a DEFAULTed column explicitly:
>     INSERT ... VALUES('foo', DEFAULT, 42);
> and being able to insert multiple explicit rows:
>     INSERT ... VALUES('foo', DEFAULT, 42), VALUES('bar', 99, 44);
> I think we should deal with all of these issues at once, which means
> it's not something to try to fix for 7.1.  Bruce, would you add a
> TODO item?
> 
>   * Bring INSERT ... VALUES up to full SQL92 spec.

Added to TODO:
 * Bring INSERT ... VALUES up to full SQL92 spec, disallow missing columns

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026

Re: AW: Shouldn't this be an error?

From
Bruce Momjian
Date:
> a DEFAULTed column explicitly:
>     INSERT ... VALUES('foo', DEFAULT, 42);
> and being able to insert multiple explicit rows:
>     INSERT ... VALUES('foo', DEFAULT, 42), VALUES('bar', 99, 44);
> I think we should deal with all of these issues at once, which means
> it's not something to try to fix for 7.1.  Bruce, would you add a
> TODO item?
> 
>   * Bring INSERT ... VALUES up to full SQL92 spec.

Oops, I have now added:

* Bring INSERT ... VALUES up to full SQL92 spec, disallow missing columns, allow DEFAULT

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
853-3000+  If your life is a hard drive,     |  830 Blythe Avenue +  Christ can be your backup.        |  Drexel Hill,
Pennsylvania19026