On Tue, Oct 20, 2026, Eugenio Zuccarelli wrote: > +- ``--summary``: If set, prints only the summary line for parsed test > + results, rather than the result of every individual test. Does nothing > + if ``--raw_output`` is set. > + > +- ``--failed``: If set, prints only the tests that failed or crashed, along > + with the summary line, once all tests have run. Does nothing if > + ``--raw_output`` is set.
Hi Eugenio, Thanks for documenting these. I compared both entries with kunit.py and kunit_parser.py and the descriptions match the code: --summary silences the per-test output and only prints the summary line, --failed prints only the failed or crashed tests plus the summary once everything has been parsed, and both are ignored with --raw_output. I noticed two things while testing the patch, in case they are useful to the KUnit maintainers: 1) With the patch applied, "make SPHINXDIRS=dev-tools htmldocs" reports two new warnings, and that file had none before: run_wrapper.rst:331: WARNING: Bullet list ends without a blank line; unexpected unindent. [docutils] run_wrapper.rst:335: WARNING: Bullet list ends without a blank line; unexpected unindent. [docutils] The continuation lines of the new items are indented by one space. Docutils wants them aligned with the text after "- ", that is two spaces, like the --json and --junit items just above. I checked that with two spaces both warnings go away, so this may be a simple fix for whoever applies the patch. 2) "Assisted-by: LLM" is not a tag that Documentation/process/ describes in docs-next, and "https://www.google.com/url?q=http://checkpatch.pl&source=gmail&ust=1793021034312000&sa=E --strict" reports it ("Unrecognized email address: 'LLM'"). Documentation/process/generated-content.rst asks for the use of a tool to be described in the changelog instead, so maybe the trailer could be dropped, or turned into a sentence in the commit message. These are only suggestions; how to proceed is of course up to you and the maintainers. Thanks, Daniel
