|From:||Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>|
|To:||David Steele <david(at)pgmasters(dot)net>|
|Cc:||Alexander Korotkov <a(dot)korotkov(at)postgrespro(dot)ru>, Tomas Vondra <tomas(dot)vondra(at)2ndquadrant(dot)com>, Magnus Hagander <magnus(at)hagander(dot)net>, Michael Paquier <michael(dot)paquier(at)gmail(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>|
|Subject:||Re: [PATCH] Atomic pgrename on Windows|
|Views:||Raw Message | Whole Thread | Download mbox | Resend email|
David Steele <david(at)pgmasters(dot)net> writes:
> On 1/11/20 5:13 PM, Alexander Korotkov wrote:
>> Regarding "pg_stat_tmp/global.stat", which is a problem in particular
>> case, we may evade file renaming altogether. Instead, we may
>> implement shared-memory counter for filename. So, instead of
>> renaming, new reads will just come to new file.
> I tend to agree with Tom on the question of portability. But it seems
> upthread we have determined that this can't be sensibly isolated into a
> Windows-specific rename() function.
> Does anyone have any further ideas? If not I feel like this patch is
> going to need to be RWF again.
So far as pg_stat_tmp is concerned, I think there is reasonable hope
that that problem is just going to go away in the near future.
I've not been paying attention to the shared-memory stats collector
thread so I'm not sure if that's anywhere near committable, but
I think that's clearly something we'll want once it's ready.
So if that's the main argument why we need this, it's a pretty
thin argument ...
regards, tom lane
|Next Message||Alvaro Herrera||2020-03-30 16:17:20||Re: ALTER TABLE ... SET STORAGE does not propagate to indexes|
|Previous Message||Pavel Stehule||2020-03-30 16:14:42||Re: materialization blocks hash join|