Hi Alvaro,
>
>> But then there's the matter of how to represent it from the grammar
>> inwards. At first I thought of creating a new parse node type, say
>> MultiResTarget that would replace ResTarget in UpdateStmt, and get it
>> from there into the optimizer and executor.
>
> Hmm, I should be passing it as a Query node, right?
>
I had (rather unsuccessfully) tried my hands on this same TODO a while
back. Here is what Tom had to say about my approach:
http://www.nabble.com/UPDATE-using-sub-selects-td9935249.html#a16108043
Also attached is the discussion on hackers to provide some more context:
http://www.nabble.com/UPDATE-using-sub-selects-td9489091.html
Might not be a big help, but just wanted to mention it here..
Regards,
Nikhils
--
http://www.enterprisedb.com