From: | Andres Freund <andres(at)anarazel(dot)de> |
---|---|
To: | Corey Huinker <corey(dot)huinker(at)gmail(dot)com> |
Cc: | "Drouvot, Bertrand" <bertranddrouvot(dot)pg(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
Subject: | Re: Generate pg_stat_get_xact*() functions with Macros |
Date: | 2023-01-05 23:21:40 |
Message-ID: | 20230105232140.ycxfh5d4lvashek2@awork3.anarazel.de |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi,
On 2023-01-05 15:19:54 -0500, Corey Huinker wrote:
> It does get me wondering, however, if we reordered the three typedefs to
> group like-typed registers together, we could make them an array with the
> names becoming defined constant index values (or keeping them via a union),
> then the typedefs effectively become:
I think that'd make it substantially enough harder to work with the
datastructures that I don't want to go there.
The "more fundamental" approach would be to switch to using a table-returning
function for accessing these stat values. When just accessing a single counter
or two, the current approach avoids the overhead of having to construct a
tuple. But after that the overhead of having to fetch the stats data (i.e. a
hash table lookup, potentially some locking) multiple times takes over.
Unfortunately there's currently no way to dynamically switch between those
behaviours.
Greetings,
Andres Freund
From | Date | Subject | |
---|---|---|---|
Next Message | Peter Geoghegan | 2023-01-05 23:27:50 | Re: BUG: Postgres 14 + vacuum_defer_cleanup_age + FOR UPDATE + UPDATE |
Previous Message | Alex Fan | 2023-01-05 23:19:55 | Re: [PATCH] Enable using llvm jitlink as an alternative llvm jit linker of old Rtdyld. |