Re: request for sql3 compliance for the update command

Bruce Momjian  writes:
> While I don't see the syntax of:
> 	update table set (col...) = ( val...)
> as valuable compared to separate col=val assignments, I do see a value
> in allowing subqueries in such assignments:
> 	update table set (col...) = ( select val ..)

Hm.  That's at least got some defensibility to it.  But does it do
anything that you can't already do with a join?

BTW, looking at the SQL99 standard, I see that you can do
UPDATE table SET ROW = foo WHERE ...

where foo is supposed to yield a row of the same rowtype as table
--- I didn't dig through the spec in detail, but I imagine foo can
be a sub-select.  I don't care a whole lot for that, though, since it
would be a real pain in the neck if you're not updating all the columns.
You'd have to go
UPDATE table SET ROW = (SELECT table.a, table.b, foo.x, ... FROM foo)

which seems ugly, tedious, and error-prone.
		regards, tom lane