Re: Add isolation test template in injection_points for wait/wakeup/detach

Hi,

On Fri, Oct 18, 2024 at 07:44:08AM +0900, Michael Paquier wrote:
> Hi all,
>
> This was on my TODO bucket for some time. The isolation test
> "inplace" in the isolation test suite for injection_points relies on a
> couple of behaviors implemented in the backend. One of them is that a
> point detach should not affect a wait. Rather than having to guess
> this stuff from the sole isolation test we have in this module, I
> would like to add a simple test to document all these assumptions.

> This also works as a simple template for a newcomer willing to
> implement an isolation test with injection points. That's less
> intimidating than the inplace test with its VACUUM and GRANT
> interactions, based on an injection point run when updating the stats
> of a relation in VACUUM.

>
> Thoughts?

I think that it makes sense to "expose" the basics in a dedicated test.

I thought that template.spec could have been another "naming" option but
basic.spec is better I think.

The test is pretty straightforward so I don't have that much to say, just one Nit:

1 ===

+# Detach before wait does not cause a wait.

What about?

Detach before wait does not cause a wait (and so wakeup produces an error)

Regards,

--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com