On 2015/09/01 12:36, Peter Geoghegan wrote:
> On Mon, Aug 31, 2015 at 8:32 PM, Satoshi Nagayasu <snaga@uptime.jp> wrote:
>> Why don't we use relation name (with looking up the catalog)
>> on query jumbling? For performance reason?
>
> I think that there is a good case for preferring this behavior. While
> it is a little confusing that pg_stat_statements does not change the
> representative query string, renaming a table does not make it a
> substantively different table.
>
> There is, IIRC, one case where a string is jumbled directly (CTE
> name). It's usually not the right thing, IMV.
>
Thanks for the comment. I've never considered that. Interesting.
From the users point of view, IMHO, it would be better to avoid
confusing if queryid is determined by only visible values -- userid,
dbid and query string itself.
BTW, I'm interested in improving the queryid portability now because
I'd like to use it in other extensions. :)
That's the reason why I'm looking at query jumbling here.
Thoughts?
Regards,
--
NAGAYASU Satoshi <snaga@uptime.jp>