Re: Linking Patchwork with Sashiko?

[Laurent Pinchart <laurent.pinchart@ideasonboard.com>]

On Mon, Jul 20, 2026 at 02:24:32PM -0400, Steven Rostedt wrote:
> On Mon, 20 Jul 2026 15:26:34 +0100 Mark Brown wrote:
> > On Mon, Jul 20, 2026 at 09:55:37AM -0400, Steven Rostedt wrote:
> > 
> > > designed. But I fortunately know of these areas and can easily see when
> > > Sashiko falls over it. Hence, these false positives do not waste my
> > > time as I'm already aware of them. (I do take a deeper look to make
> > > sure it's not reporting something similar that may truly be a bug).  
> > 
> > OTOH I had a submitter up to v14 on a new series over the weekend since
> > they were iterating Sashiko feedback.
> 
> Hmm, we should state that Sashiko is sometimes wrong.  A submitter to
> me asked what they should do about a Sashiko feedback they diddn't
> understand, and after reading the Sashiko review, I told them to ignore
> it.

We discussed this at length in linux-media. There were proposals for a
custom header telling patch authors that in case they think a flagged
issue may be incorrect they should reach to an experienced reviewer (or
just ignore it, the wording went through multiple iterations). The
problem there is that, by definition, a junior, unexperienced developer
will have more trouble determining if feedback is correct or not.

> > > Do we need to add an flag to the MAINTAINERS file that states which
> > > sub-systems uses AI?  
> > 
> > That's probably too complicated - some poeple will want submitters to do
> > things like ask submitters to address Sashiko reviews before the
> > maintainers will look at it, other subsystems might make use of Sashiko
> > reviews or whatever but not make it a requirement for submitters to do
> > so.
> 
> Is it too complicated? Do you mean it's too complicated if one
> sub-system says "resolve all Sashiko feedback you receive" and another
> is "You may ignore it"?
> 
> We probably don't want maintainers to require resolving all Sashiko
> reviews, but at least respond to them. Where "I don't know" and "I
> think this review is wrong" are legitimate responses. It is basically
> saying "I want to speak to a person" when drudging through the phone
> automated AI teleprompters.

-- 
Regards,

Laurent Pinchart