From: | Dimitri Fontaine <dfontaine(at)hi-media(dot)com> |
---|---|
To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
Cc: | pgsql-hackers(at)postgreSQL(dot)org |
Subject: | Re: EXPLAIN, utility statement parameters, and recent plpgsql changes |
Date: | 2010-01-14 20:17:57 |
Message-ID: | m2tyuoxwy2.fsf@hi-media.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
> Dimitri Fontaine <dfontaine(at)hi-media(dot)com> writes:
>> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
>>> This works well enough for regular DML statements, but it falls down for
>>> EXPLAIN which is a utility statement, because *parse analysis of utility
>>> statements doesn't do anything*. EXPLAIN actually does the parse
>>> analysis of its contained statement at the beginning of execution.
>>> And that is too late, in the scenario Pavel exhibited. Why is it too
>>> late? Because SPI_cursor_open_internal() intentionally "freezes" the
>>> ParamListInfo struct after doing initial parsing: what it copies into
>>> the cursor portal is just a static list of data values without the
>>> parser hooks (see copyParamList).
>
>> Would it make any sense for this function to get to call the hook in the
>> case a utility statement is being processed?
>
> Well, the point of the hook is to change the results of parse
> transformation, so just calling it doesn't do much --- you have to apply
> the whole parse analysis process, *and keep the resulting tree*.
Could that be done in the function, in the phase you call "doing initial
parsing"?
--
dim
From | Date | Subject | |
---|---|---|---|
Next Message | Kevin Grittner | 2010-01-14 20:23:01 | Re: Testing with concurrent sessions |
Previous Message | Robert Haas | 2010-01-14 20:13:46 | Re: Streaming replication, retrying from archive |