Re: pg_stat_statements: calls under-estimation propagation

On Fri, Oct 11, 2013 at 2:49 AM, Alvaro Herrera
<alvherre(at)2ndquadrant(dot)com> wrote:
> Daniel Farina escribió:
>> On Thu, Oct 10, 2013 at 7:40 AM, Fujii Masao <masao(dot)fujii(at)gmail(dot)com> wrote:
>
>> > In my test, I found that pg_stat_statements.total_time always indicates a zero.
>> > I guess that the patch might handle pg_stat_statements.total_time wrongly.
>> >
>> > + values[i++] = DatumGetTimestamp(
>> > + instr_get_timestamptz(pgss->session_start));
>> > + values[i++] = DatumGetTimestamp(
>> > + instr_get_timestamptz(entry->introduced));
>> >
>> > These should be executed only when detected_version >= PGSS_TUP_LATEST?
>>
>> Yes. Oversight.
>
> Hmm, shouldn't this be conditional on a new PGSS_TUP_V1_2?

I was just thinking the same thing. Agreed.

Regards,

--
Fujii Masao