Re: [PATCH v4 2/4] dt-bindings: hwmon: Add Sensirion SHT30 series

[Roman Gushchin]

Hello,

The pseudo-commit message is intentional, Chris Mason initially
prompted it, and I'm honestly just using his format. The intention
here is to ensure the LLM reviews the correct commit and at least
understands what it's doing. Idk if it's really useful with Sashiko,
but it's there and hopefully not too distracting.

Re previous versions: it's not implemented yet, but I plan to add this
eventually.

Thanks!


On Thu, Mar 26, 2026 at 5:03 PM Conor Dooley <conor@xxxxxxxxxx> wrote:
>
> On Thu, Mar 26, 2026 at 04:55:39PM -0700, Guenter Roeck wrote:
> > On 3/26/26 10:45, Conor Dooley wrote:
> > > On Thu, Mar 26, 2026 at 05:42:30PM +0000, Conor Dooley wrote:
> > > > On Wed, Mar 25, 2026 at 06:05:22PM -0700, Guenter Roeck wrote:
> > > > > On 3/25/26 11:20, Conor Dooley wrote:
> > > > > > On Wed, Mar 25, 2026 at 05:08:08PM +0800, Zaixiang Xu wrote:
> > > > > > > Add YAML devicetree binding schema for Sensirion SHT30 series.
> > > > > > > Use fallback compatibles for compatible chips and add optional
> > > > > > > interrupts and vdd-supply properties.
> > > > > > >
> > > > > > > Reported-by: kernel test robot <lkp@xxxxxxxxx>
> > > > > > > Closes: https://lore.kernel.org/r/202603212044.BRPaiz86-lkp@xxxxxxxxx/
> > > > > >
> > > > > > The robot did not report that this binding was missing.
> > > > > > It also told you not to add these tags.
> > > > > >
> > > > > > You also ignored my and Krzysztof's reviews.
> > > > > >
> > > > > > NAK.
> > > > > >
> > > > >
> > > > > Maybe we should just point to AI feedback:
> > > > >
> > > > > https://sashiko.dev/#/patchset/1774429690-129139-1-git-send-email-zaixiang.xu.dev%40gmail.com
> > > > >
> > > > > and only get involved after AI does not report any problems.
> > > > >
> > > >
> > > > The presentation of info in that is weird, it creates a pseudo-commit
> > > > message, and then goes on to talk about things that the pseudo-commit
> > > > message has had culled.
> > >
> > > How good is this LLM stuff at figuring out if previous review feedback
> > > has been resolved? Or is it not capable of looking at earlier revisions?
> >
> >
> > Both are interesting questions. For the first one, I had noticed that, but
> > to me it seemed that the pseudo-commit message is an attempt to create a concise
> > version of the commit message, and I have not seen evidence that the later
> > comments did not also apply to the real commit message.
>
> Yeah, the comments did apply to the real commit message. I was just
> confusing to me to see what looked to me like a commit message not
> contain the info that was being referred to later on. I thought that the
> LLM was hallucinating the comments until I went back to this mail and
> realised the text was here.
>
> >
> > Anyway, copying Roman for feedback on both.
> >
> > Thanks,
> > Guenter
> >