Re: [PATCH 2/2] format-patch: allow forcing the use of in-body From: header
Hi Junio,
On Fri, 26 Aug 2022, Junio C Hamano wrote:
Show 14 quoted lines
> Users may be authoring and committing their commits under the same
> e-mail address they use to send their patches from, in which case
> they shouldn't need to use the in-body From: line in their outgoing
> e-mails. At the receiving end, "git am" will use the address on the
> "From:" header of the incoming e-mail and all should be well.
>
> Some mailing lists, however, mangle the From: address from what the
> original sender had; in such an unfortunate situation, the user may
> want to add the in-body "From:" header even for their own patch.
>
> "git format-patch --[no-]force-inbody-from" was invented for such
> users.
>
> Note. This is an uncooked early draft.
Did you mean to mark the patch as [RFC], then?
Show 8 quoted lines
> Things to think about include (but not limited to, of course):
>
> * Should this rather be --use-inbody-from=yes,no,auto tristate,
> that defaults to "auto", which is the current behaviour i.e.
> "when --from is given, add it only when it does not match the
> payload". "yes" would mean "always emit the --from address as
> in-body From:" and "no" would mean ... what? "Ignore --from"?
> Then why is the user giving --from in the first place?
I would offer up the suggestion `--in-body-from={never,always,auto}` for consideration.
> * Should it be "inbody" or "in-body"?
> * Should it have a corresponding configuration variable?
Probably. The commit message talks about mailing lists requiring different behavior from the default, which is likely to affect all patches generated from a corresponding local checkout. Having a config variable would lower the cognitive burden of having to remember this process detail.
> * Should this patch be scrapped and the feature should be done
> inside "git send-email" instead?
Since it affects the `--pretty=email` mode, the current patch seems to aim for the correct layer.
Show 10 quoted lines
> diff --git a/builtin/log.c b/builtin/log.c
> index 9b937d59b8..83b2d01b49 100644
> --- a/builtin/log.c
> +++ b/builtin/log.c
> @@ -1897,6 +1897,8 @@ int cmd_format_patch(int argc, const char **argv, const char *prefix)
> N_("show changes against <refspec> in cover letter or single patch")),
> OPT_INTEGER(0, "creation-factor", &creation_factor,
> N_("percentage by which creation is weighted")),
> + OPT_BOOL(0, "force-inbody-from", &rev.force_inbody_from,
> + N_("Use in-body From: even for your own commit")),Please start the usage text in lower-case, to keep it consistent with the rest of the usage texts.
Also, I would like to avoid the personal address "you" in that text, and also the verb "use". Maybe something like this:
show in-body From: even if identical to the header
Show 13 quoted lines
> OPT_END()
> };
>
> diff --git a/pretty.c b/pretty.c
> index 51e3fa5736..e266208c0b 100644
> --- a/pretty.c
> +++ b/pretty.c
> @@ -483,6 +483,8 @@ static int use_inbody_from(const struct pretty_print_context *pp, const struct i
> return 0;
> if (ident_cmp(pp->from_ident, ident))
> return 1;
> + if (pp->rev && pp->rev->force_inbody_from)
> + return 1;
It would probably make sense to move this before `ident_cmp()`, to avoid unneeded calls ("is the ident the same? no? well, thank you for your answer but I'll insert the header anyway!").
Show 12 quoted lines
> return 0;
> }
>
> diff --git a/revision.h b/revision.h
> index bb91e7ed91..a2d3813a21 100644
> --- a/revision.h
> +++ b/revision.h
> @@ -208,6 +208,7 @@ struct rev_info {
>
> /* Format info */
> int show_notes;
> + unsigned int force_inbody_from;The reason why this isn't added to the `:1` bits below is probably the anticipation of the tri-state, but if that tri-state never materializes, adding it as a bit is still the right thing to do.
Show 23 quoted lines
> unsigned int shown_one:1,
> shown_dashes:1,
> show_merge:1,
> diff --git a/t/t4014-format-patch.sh b/t/t4014-format-patch.sh
> index fbec8ad2ef..a4ecd433e2 100755
> --- a/t/t4014-format-patch.sh
> +++ b/t/t4014-format-patch.sh
> @@ -1400,6 +1400,19 @@ test_expect_success '--from omits redundant in-body header' '
> test_cmp expect patch.head
> '
>
> +test_expect_success 'with --force-inbody-from, --from keeps redundant in-body header' '
> + git format-patch --force-inbody-from \
> + -1 --stdout --from="A U Thor <author@example.com>" >patch &&
> + cat >expect <<-\EOF &&
> + From: A U Thor <author@example.com>
> +
> + From: A U Thor <author@example.com>
> +
> + EOF
> + sed -ne "/^From:/p; /^$/p; /^---$/q" patch >patch.head &&
> + test_cmp expect patch.head
> +'
The test script starts to look a bit non-DRY with all those repetitions of `A U Thor <author@example.com>`, but that's hardly the responsibility of this here patch to address.
Thank you, Dscho
Show 8 quoted lines
> +
> test_expect_success 'in-body headers trigger content encoding' '
> test_env GIT_AUTHOR_NAME="éxötìc" test_commit exotic &&
> test_when_finished "git reset --hard HEAD^" &&
> --
> 2.37.2-587-g47adba97a9
>
>
Junio C Hamano
