Re: [PATCH b4] send: Add --force-cover-letter to send separate cover for single patch patch series

["Antonin Godard" <antonin.godard@bootlin.com>]

Hi,

On Tue Jan 27, 2026 at 1:53 PM CET, Breno Leitao wrote:
> By default, b4 mixes the cover letter content into the patch body
> (under the --- cut line) when sending a single-patch series. Add
> --force-cover-letter option to allow sending a separate cover letter
> message (as 0/1) even for single-patch series.
>
> This is useful when developers want to send a long cover letter with
> extra information about an issue, and a commit. Having the cover-letter
> "mixed-in" makes it hard to follow.
>
> Example:
>  * https://lore.kernel.org/all/20260127-ipmi-v1-0-ba5cc90f516f@debian.org/
>
> Signed-off-by: Breno Leitao <leitao@debian.org>
> ---
>  src/b4/command.py |  2 ++
>  src/b4/ez.py      | 16 +++++++++-------
>  2 files changed, 11 insertions(+), 7 deletions(-)
>
> diff --git a/src/b4/command.py b/src/b4/command.py
> index 1f8b8f1..0e685cd 100644
> --- a/src/b4/command.py
> +++ b/src/b4/command.py
> @@ -397,6 +397,8 @@ def setup_parser() -> argparse.ArgumentParser:
>                           help='Resend a previously sent version of the series')
>      sp_send.add_argument('--no-sign', action='store_true', default=False,
>                           help='Do not add the cryptographic attestation signature header')
> +    sp_send.add_argument('--force-cover-letter', action='store_true', default=False,
> +                         help='Send a cover letter even for single-patch series')

I've had a different idea with this which would go a bit further.

Instead of --force-cover-letter, have a --cover-letter option which acts as
follows:

* --cover-letter=yes: always generate a cover-letter, no matter if it's a single
  patch or not

* --cover-letter=auto (default): generate a cover-letter when there is more
  than one patch. This is the current behavior.

* --cover-letter=no: do not generate a cover-letter, even on multi-patch series

This would be backed up by a setting (b4.send-cover-letter?) which can be set
globally or on a per-branch basis.

I don't know how easy it would be to support the last use-case though, as I
don't know how b4 is going to behave with a multi-patch series with no
cover-letter, but since you figured out the code paths I guess it would be
relatively easy to try? Obviously, --edit-cover would have to be adapted as well
to refuse editing the cover when --cover-letter=no.

I do find myself needing the last use-case sometimes, when the series does
not require a cover-letter.

What do you think?

Antonin