Re: Only one version can be installed when using extension_control_path

Michael Paquier <michael@paquier.xyz>

From: Michael Paquier <michael@paquier.xyz>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: "David G. Johnston" <david.g.johnston@gmail.com>, Peter Eisentraut <peter@eisentraut.org>, Matheus Alcantara <matheusssilv97@gmail.com>, Pierrick <pierrick.chovelon@dalibo.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2025-09-10T23:38:44Z
Lists: pgsql-hackers
On Mon, Sep 08, 2025 at 11:35:42AM -0400, Tom Lane wrote:
> 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.

It would be too late once we are in GA, and we are not yet in GA.  At
this point I think that this should be Peter's call.

I am not in his shoes, still I suspect like you that we may have not
seen the end of it yet.  If we're confident enough that we can fix
anything, that's fine as a response to me, but discussions are also
involving how available extensions are seen at catalog level.  This
bit is concerning for me, but I'm also of a pessimistic nature.
--
Michael

Commits

  1. Hide duplicate names from extension views