Re: Only one version can be installed when using extension_control_path

From: Pierrick <pierrick(dot)chovelon(at)dalibo(dot)com>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, "David G(dot) Johnston" <david(dot)g(dot)johnston(at)gmail(dot)com>
Cc: Peter Eisentraut <peter(at)eisentraut(dot)org>, Matheus Alcantara <matheusssilv97(at)gmail(dot)com>, "pgsql-hackers(at)lists(dot)postgresql(dot)org" <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Only one version can be installed when using extension_control_path
Date: 2025-09-09 07:53:44
Message-ID: c5d80de9-8b63-45f8-8147-deeccf4b0198@dalibo.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers


On 08/09/2025 17:35, Tom Lane wrote:
> "David G. Johnston" <david(dot)g(dot)johnston(at)gmail(dot)com> writes:
>> On Monday, September 8, 2025, Matheus Alcantara <matheusssilv97(at)gmail(dot)com>
>> wrote:
>>> On this step it will search the .control
>>> file on paths at extension_control_path in order and it will use the
>>> first one that it finds and based on the .control file found it will
>>> install the extension using the version specified on the command.
>> Then pg_available_extensions seems broken - it lists extensions as
>> available that are not accessible to the user due to this policy. Maybe we
>> need to add something indicating that certain rows are hidden behind the
>> present path setting which would need to be changed if one wishes to
>> install them. Also seems like the extension location should be part of the
>> output too.

I agree. If an extension is the list, I expect to be able to install it.

Either :
- we hide the latest version found in paths in pg_available_extensions ;
- we give the possibility to install all listed versions.
The second option seems more relevant to me.

Adding extension location is a good idea.

> Between this and previously-identified problems (commits 81eaaa2c4,
> f777d7738), it seems clear that extension_control_path (which is a new
> thing in v18) was very poorly thought out. I wonder if it's too late
> to revert it so that we can redesign it more carefully.
>
> regards, tom lane

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Fujii Masao 2025-09-09 07:56:15 Re: pg_restore --no-policies should not restore policies' comment
Previous Message Chao Li 2025-09-09 07:52:54 Re: Fix missing EvalPlanQual recheck for TID scans