Re: Patch pre-submission to Sashiko

["Theodore Tso" <tytso@mit.edu>]

On Mon, Oct 19, 2026 at 11:53:27AM -0500, Shuah Khan wrote:
> 
> When a submitter uses LLMs to generate a patch, it is reasonable for
> maintainers and reviewers to ask them to use shashiko to find problems
> before sending it to us. Asking for a shashkiko run before submitting
> patches could become part of the patch submitting process.
> 
> How about updating the submitting patches document to add a section on
> this with instructions on how to run it?
> 
> This shifts the responsibility to run shashiko to people using LLMs.

If people are using LLM's to generate a patch, it would be useful if
we can make it super easy for LLM's to call out to Sashiko as an
sub-agent.  We would probably want to make it easier for LLM's to run
our kunit and selftests as well.  Basically, we want to push the work
to the submmiter, instead of expecting the maintainer and/or reviewers
to do that work --- especially since if it's in the agenting coding
loop, the LLM agent can fix the problem more quickly and without
burning human review time.

					- Ted