Alvaro Herrera <alvherre@alvh.no-ip.org> writes:
> David Steele wrote:
>> Based on Tom's feedback, and hearing no opinions to the contrary, I have
>> marked this patch Rejected.
> I think I opine contrarywise, but I haven't made time to review the
> status of this in detail. I'm fine with keeping it rejected for now,
> but I reserve the option to revive it in the future.
I'm fine with reviving it if someone can find a way around the new-
reserved-word problem. But that's gonna be a bit hard given that
the patch wants to do
database_name:
ColId
| CURRENT_DATABASE
You might be able to preserve the accessibility of the current_database()
function by making CURRENT_DATABASE into a type_func_name_keyword instead
of a fully-reserved word. But that's ugly (I think you'd need a single-
purpose production for it to be used as a function), and you've still
broken any SQL code using "current_database" as a table or column name.
I'm dubious that the remaining use-case for the feature is worth it.
regards, tom lane