| From: | Michael Paquier <michael(at)paquier(dot)xyz> |
|---|---|
| To: | Daniel Gustafsson <daniel(at)yesql(dot)se> |
| Cc: | Corey Huinker <corey(dot)huinker(at)gmail(dot)com>, Mike Fiedler <miketheman(at)gmail(dot)com>, Postgres hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: Emit namespace in post-copy output |
| Date: | 2021-07-29 00:57:30 |
| Message-ID: | YQH8+rqjOkoaY3Ia@paquier.xyz |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, Jul 28, 2021 at 04:15:19PM +0200, Daniel Gustafsson wrote:
> Since get_namespace_name() returns a palloced string, this will lead to a 2x
> leak of the namespace length as opposed to the 1x of today. While hardly a big
> deal, it seems prudent to cap this by storing the returned string locally now
> that we need it twice.
I don't think this matters much. A quick read of the code shows that
this memory should be allocated within the transaction context running
CLUSTER/VACUUM FULL, and that gets free'd in the internal calls of
CommitTransactionCommand().
--
Michael
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Ranier Vilela | 2021-07-29 01:03:29 | Re: Out-of-memory error reports in libpq |
| Previous Message | Michael Paquier | 2021-07-29 00:52:08 | Re: Use WaitLatch for {pre, post}_auth_delay instead of pg_usleep |