On 5/14/19 8:34 AM, Joe Conway wrote:
> On 5/13/19 8:32 PM, Michael Paquier wrote:
>> On Mon, May 13, 2019 at 01:06:10PM -0700, legrand legrand wrote:
>>> that finished commited
>>> "pgsql: Add hooks for session start and session end"
>>>
https://www.postgresql.org/message-id/flat/575d6fa2-78d0-4456-8600-302fc35b2591%40dunslane.net#0819e315c6e44c49a36c69080cab644d
>>>
>>> but was finally rollbacked because it didn't pass installcheck test
>>> ...
>>>
>>> Maybe you are able to patch your pg installation,
>>> it would be a solution of choice (there is a nice exemple
>>> of extension included)
>> You will need to patch Postgres to add this hook, and you could
>> basically reuse the patch which has been committed once. I don't
>> think that it would be that much amount of work to get it working
>> correctly on the test side to be honest, so we may be able to get
>> something into v13 at this stage. This is mainly a matter of
>> resources though, and of folks willing to actually push for it.
>
> I am interested in this, so if Andrew wants to create a buildfarm module
> I will either add it to rhinoceros or stand up another buildfarm animal
> for it. I am also happy to help push it for v13.
>
I've just been looking at this again. I think the right way to test it
is not to use the regression framework but to use a TAP test that would
preload the test module. Then it would just happen as part of all the
other TAP tests with no extra buildfarm module needed.
cheers
andrew
--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
