Re: [PATCH] docs: submitting-patches: Require base tree information

[Krzysztof Kozlowski <krzk@kernel.org>]

On 27/10/2026 13:29, Laurent Pinchart wrote:
> On Mon, Oct 26, 2026 at 11:48:16AM +0100, Krzysztof Kozlowski wrote:
>> On 26/10/2026 11:40, Laurent Pinchart wrote:
>>> On Thu, Oct 22, 2026 at 08:57:55AM +0200, Krzysztof Kozlowski wrote:
>>>> That's valid and for 95% (guessing the number...) of cases will be
>>>> solved by having linux-next as a base, thus requiring the base
>>>> information is redundant. It's just linux-next.
>>>>
>>>> When patchset does NOT apply on linux-next or maintainer tree is
>>>> non-obvious, the base should be provided, but that is only "if ->
>>>> should". I think the proposal here is a bit different.
>>>
>>> What's the downside though ? As git can easily provide the base
>>> information once configured to do so, what's the downside of asking
>>> contributors to include it ?
>>
>> For me the downside is that if you require it, half or more of my
>> patches would be rejected
> 
> I don't think we want to automate rejection of patches that provide no
> base information, and certainly not as a first step. The goal as I see
> it is to encourage people to provide the information. It's a process
> that will take time.

We already encourage that in current wording and b4 automatically
provides that.

If you want to encourage, do not make it a requirement "with only rare
exceptions".


> 
>> and I would need to add unnecessary step to
>> remember.
> 
> It doesn't have to require manual work every time you send paches.
> format.useAutoBase in conjunction with specifying upstreams of branches
> helps.

But then base would point nowhere... Upstream of my branches are my
github branches, so I always know what is up2date against my github.
Otherwise how do you track - with multi-machine and fully backed up
repos - whether your local changes are updated against your backed up
branch on github/forge?


> 
>> Since this step is pointless - it would be linux-next
>> (assuming in the first place that this is what the author wanted here!)
>> - it is a waste of time for me and potential resubmit of series, just to
>> fulfill requirement which brings no benefits. So I have downside and I
>> see no upside for my case, when you require the base.
>>
>> This might be affecting others as well, but now speaking about my
>> workflow - I have a lot of independent cleanups, which I stack on top of
>> each other, sometimes series, sometimes not, but stacking is important
>> so when I run `make W=1` it passes.
>>
>> What's more important, I reject the entire reasoning/cause of this
>> change, which claimed you cannot review and therefore accept such
>> patches. It's simply not true.
> 
> That's not my claim.

That's the claim from commit msg. And wording should rather be "which
claimed one cannot review".

Best regards,
Krzysztof