On Fri, Jul 20, 2012 at 1:31 AM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> I wrote:
>> I think the way to solve this is to do whatever it takes to get access
>> to the subplan targetlist. We could then do something a bit cleaner
>> than what the named-rowtype code is currently doing: if there are
>> resjunk columns in the subplan targetlist, use the tlist to create a
>> JunkFilter, and then pass the tuples through that. After that we can
>> insist that the tuples don't have any extra columns.
>
> Here's a draft patch for that. It wasn't quite as ugly as I feared.
> A lot of the apparent bulk of the patch is because I chose to split
> ExecEvalVar into separate functions for the scalar and whole-row
> cases, which seemed appropriate because they now get different
> ExprState node types.
Thanks for that! Applying the patch and confirming the fix turned up
no issues. I did a perfunctory review and it all looks pretty good:
maybe ExecInitExpr could use a comment describing the
InvalidAttrNumber check though...it's somewhat common knowledge that
InvalidAttrNumber means row variables but it's also used to initialize
variables before loops scans and things like that.
merlin