Re: [PATCH] Documentation: coding-assistants: add optional Acted-By: trailer

[Blake]

On Tuesday, 21 April 2026 at 3:40 AM, Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> I really do not understand, how would this actually be used?

Git history uses email addresses as the de-facto identity key and
email addresses aren't stable across a lifetime or a career and,
increasingly, aren't meaningful even within a single commit.
Contributors move between employers and domains over time; companies
rename and migrate their workforce onto new domains; and plenty of
commits land with From: lines like blake@desktop or
root@localhost.localdomain because git's username@hostname fallback
fired before user.email was set.

.mailmap partially fixes the first case inside a single tree but
doesn't travel across repositories and can't synthesise an identity
where the From: never carried one. A declared Acted-By trailer gives
the human a stable key that survives employer changes, domain
migrations, misconfigured clients, and cross-tree lookups.

> And as you have to have a signed-off-by, why would you use two
> different names for yourself this way?

Signed-off-by and Acted-By answer different questions. Signed-off-by
requires a real name for DCO attestation. Acted-By names the stable
working identity the same human operates under across contexts —
which for most contributors on a stable setup collapses to the same
name, and the trailer adds nothing. They diverge when:

  - the same human contributes under different employer emails over
    a long career;
  - an employer renames and migrates its workforce to a new domain;
  - a contributor publishes under a stable pseudonym but their DCO
    Signed-off-by has to carry their legal name;
  - the From: line is machine-generated noise (blake@desktop,
    user@hostname) and carries no identity information at all.

> And who is asking for this?  Who would want to use it?  How would
> you use it for this very commit?

For this very commit I wouldn't need it as Signed-off-by is
sufficient. The patch defines a slot in the docs; the commits that
benefit from the slot are other commits in other trees where the
divergence cases above bite. Pseudonymous contributors and
long-career maintainers whose email addresses have changed are the
current constituency; anyone writing provenance or review-routing
tooling that today falls back on email-fuzzing is the other.

If the sense is that optional trailers with narrow audiences don't
earn a footnote in coding-assistants.rst, I'm happy to withdraw and
keep this out-of-tree.

thanks,
Blake