From: | Andrew Dunstan <andrew(at)dunslane(dot)net> |
---|---|
To: | Jehan-Guillaume de Rorthais <jgdr(at)dalibo(dot)com> |
Cc: | Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, Mark Dilger <mark(dot)dilger(at)enterprisedb(dot)com>, PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: multi-install PostgresNode fails with older postgres versions |
Date: | 2021-04-19 12:11:01 |
Message-ID: | a80421c0-3d7e-def1-bcfe-24777f15e344@dunslane.net |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On 4/17/21 12:35 PM, Andrew Dunstan wrote:
>
>> OK, here is more WIP on this item. I have drawn substantially on Mark's
>> and Jehan-Guillaime's work, but it doesn't really resemble either, and I
>> take full responsibility for it.
>>
>> The guiding principles have been:
>>
>> . avoid doing version tests, or capability tests which are the moral
>> equivalent, and rely instead on pure overloading.
>>
>> . avoid overriding large pieces of code.
>>
>>
>> The last has involved breaking up some large subroutines (like init)
>> into pieces which can more sensibly be overridden. Breaking them up
>> isn't a bad thing to do anyway.
>>
>> There is a new PostgresVersion object, but it's deliberately very
>> minimal. Since we're not doing version tests we don't need more complex
>> routines.
>>
>>
>
> I should have also attached my test program - here it is. Also, I have
> been testing with the binaries which I've published here:
> <https://gitlab.com/adunstan/pg-old-bin> along with some saved by my
> buildfarm animal.
>
>
I've been thinking about this some more over the weekend. I'm not really
happy with any of the three approaches to this problem:
a) Use version or capability tests in the main package
b) No changes to main package, use overrides
c) Changes to main package to allow smaller overrides
The problem is that a) and c) involve substantial changes to the main
PostgresNode package, while b) involves overriding large functions (like
init) sometimes for quite trivial changes.
I think therefore I'm inclined for now to do nothing for old version
compatibility. I would commit the fix for the IPC::Run caching glitch,
and version detection. I would add a warning if the module is used with
a version <= 11.
The original goal of these changes was to allow testing of combinations
of different builds with openssl and nss, which doesn't involve old
version compatibility.
As far as I know, without any compatibility changes the module is fully
compatible with releases 13 and 12, and with releases 11 and 10 so long
as you don't want a standby, and with releases 9.6 and 9.5 if you also
don't want a backup. That makes it suitable for a lot of testing without
any attempt at version compatibility.
We can revisit compatibility further in the next release.
cheers
andrew
--
Andrew Dunstan
EDB: https://www.enterprisedb.com
From | Date | Subject | |
---|---|---|---|
Next Message | Amit Kapila | 2021-04-19 12:22:47 | Re: Forget close an open relation in ReorderBufferProcessTXN() |
Previous Message | Andrei Zubkov | 2021-04-19 12:03:27 | Re: [PATCH] Tracking statements entry timestamp in pg_stat_statements |