On 11/16/23 15:26, Mark Brown wrote:
On Thu, Nov 16, 2023 at 02:14:24PM +0200, Nikolai Kondrashov wrote:
Yes, that would be better indeed. However, checkpatch.pl doesn't process cover letters, and so we would have no automated way to advertise and nudge people towards testing.
Back when I used to run checkpatch it seemed to cope, it obviously wouldn't find much to look at in there but you could feed it an entire series with cover letter and the cover letter wouldn't cause any extra issues.
Ah, good to know, thank you. The question now is whether we can expect, or require, submitters to run checkpatch.pl on the complete patchset, cover letter included, *before* sending it.
P.S. Git forges do that for you by nature of actually running the tests themselves, automatically. *Ducks*
The ability of forges to run tests is not hugely relevant to large portions of the kernel, for drivers you're wanting the tests to run on the relevant hardware and even core changes will often need hardware that exercises the relevant features to run. In these areas you're more just looking for someone to say that they've done relevant testing but there's not a substantial difference between say a comment on a pull request or a followup email.
Agreed.
Still, there *are* largely hardware-independent (and thus maybe more impactful) parts of the kernel too.
Plus, you know yourself there's a bunch of companies willing (if not outright clamoring) to contribute their machine time for testing, provided some good will and authentication on the part of the contributors. And git forges provide good support for the latter. So perhaps *some* special hardware *could* be arranged there too, making it more useful.
Nick