|From:||Julien Rouhaud <rjuju123(at)gmail(dot)com>|
|To:||Magnus Hagander <magnus(at)hagander(dot)net>|
|Cc:||Sergei Kornilov <sk(at)zsrv(dot)org>, legrand legrand <legrand_legrand(at)hotmail(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org>|
|Subject:||Re: pg_stat_statements oddity with track = all|
|Views:||Raw Message | Whole Thread | Download mbox | Resend email|
On Thu, Apr 08, 2021 at 10:30:53AM +0200, Magnus Hagander wrote:
> I agree. If those numbers are indeed representable, it seems like
> better to pay that overhead than to pay the overhead of trying to
> de-dupe it.
> Let's hope they are :)
> Looking through ti again my feeling said the toplevel column should go
> after the queryid and not before, but I'm not going to open up a
> bikeshed over that.
> I've added in a comment to cover that one that you removed (if you did
> send an updated patch as you said, then I missed it -- sorry), and
> applied the rest.
Oops, somehow I totally forgot to send the new patch, sorry :(
While looking at the patch, I unfortunately just realize that I unnecessarily
bumped the version to 1.10, as 1.9 was already new as of pg14. Honestly I have
no idea why I used 1.10 at that time. Version numbers are not a scarce
resource but maybe it would be better to keep 1.10 for a future major postgres
If yes, PFA a patch to merge 1.10 in 1.9.
|Next Message||Michael Paquier||2021-04-08 12:05:22||Re: PostgreSQL 14 Feature Freeze + Release Management Team (RMT)|
|Previous Message||David Rowley||2021-04-08 12:00:59||Re: Binary search in ScalarArrayOpExpr for OR'd constant arrays|