On Thu, Jul 20, 2023 at 02:37:46PM -0700, Jakub Kicinski wrote:

On Thu, 20 Jul 2023 16:15:26 +0100 Conor Dooley wrote:

quoted

..I noticed that none of these sections address actually testing the
code they're responsible for on a (semi-)regular basis. Sure, that comes
as part of reviewing the patches for their code, but changes to other
subsystems that a driver/feature maintainer probably would not have been
CCed on may cause problems for the code they maintain.
If we are adding a doc about best-practice for maintainers, I think we
should be encouraging people to test regularly.

I think our testing story is too shaky to make that a requirement.
Differently put - I was never able to get good upstream testing running
when I worked for a vendor myself so I wouldn't know how to draw 
the lines.

I'm not saying it needs to be added as a must level item, some words to the
effect of
  Maintainers should test the drivers/features they are responsible for on a
  regular basis, independent of patches that modify their area of
  responsibility. This helps ensure that changes to other parts of the kernel
  do not introduce regressions in their driver/feature."
would suffice IMO.

The doc as it is is a useful addition though, so you can add a
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
if you like.

Thanks,
Conor.