Bruce Momjian <bruce@momjian.us> writes:
> Tom Lane wrote:
>> Hm, you're right. This arises from the fact that _SPI_execute_plan
>> rejects cursor-related utility statements. While I'd never stopped
>> to question that before, it does seem like this restriction is a
>> bit pointless. Does anyone remember why it's like that?
> Is there anything to do on this item?
I dug in the archives and realized that SPI was originally written by
Vadim, not Jan as I'd been thinking, so there's nobody left on the
project who has any special insight into this. I found this message
describing it:
http://archives.postgresql.org/pgsql-hackers/1997-08/msg00338.php
wherein Vadim says
>>> 4. I have some code for server programming interface (SPI)
>>> which allows to run queries from user defined C-functions.
>>> With current postgres limitations (no nested transactions, cursors
>>> inside BEGIN/END only) SPI disallows using of BEGIN/END & cursors.
>>> It's bad for procedures but it's enough for triggers!
so it seems he saw this just as an implementation restriction
rather than a fundamental property of SPI. And I don't see
why cursors being within-transaction only means SPI shouldn't
touch them --- maybe he just wasn't thinking carefully about that.
I'd venture that we should try to get rid of the restriction, but I'm
unsure whether removing the error check is sufficient or whether there
are real problems it's preventing.
regards, tom lane