Question on range-diff and notes.displayref

Question on range-diff and notes.displayref

[Elijah Newren]

Should git notes show up in a range-diff?  I happened to have
    notes.displayref=refs/notes/amlog
set in my git.git repo, and saw the below in my range-diff:

On Tue, Jul 31, 2018 at 10:12 AM, Elijah Newren <newren@gmail.com> wrote:
> 1:  4a1c9c3368 ! 1:  00f94a8b41 t1015: demonstrate directory/file conflict recovery failures
>     @@ -14,7 +14,6 @@
>
>          Signed-off-by: Elijah Newren <newren@gmail.com>
>          Signed-off-by: Junio C Hamano <gitster@pobox.com>
>     -    Message-Id: <20180713163331.22446-2-newren@gmail.com>
>
>      diff --git a/t/t1015-read-index-unmerged.sh b/t/t1015-read-index-unmerged.sh
>      new file mode 100755
> 2:  e105e8bfbd ! 2:  d3b8d7edb6 read-cache: fix directory/file conflict handling in read_index_unmerged()
>     @@ -59,7 +59,6 @@
>
>          Signed-off-by: Elijah Newren <newren@gmail.com>
>          Signed-off-by: Junio C Hamano <gitster@pobox.com>
>     -    Message-Id: <20180713163331.22446-3-newren@gmail.com>
>
<snip>


Maybe this is expected or wanted (tbdiff also shows the git notes for
what it's worth), but it seemed somewhat surprising to me.  I'd rather
not see such "differences" displayed for the patch series that I'm
submitting, but perhaps others see it differently?

Elijah

Re: Question on range-diff and notes.displayref

[Junio C Hamano]

Elijah Newren <newren@gmail.com> writes:

> Should git notes show up in a range-diff?  I happened to have
>     notes.displayref=refs/notes/amlog
> set in my git.git repo, and saw the below in my range-diff:
>
> On Tue, Jul 31, 2018 at 10:12 AM, Elijah Newren <newren@gmail.com> wrote:
>> 1:  4a1c9c3368 ! 1:  00f94a8b41 t1015: demonstrate directory/file conflict recovery failures
>>     @@ -14,7 +14,6 @@
>>
>>          Signed-off-by: Elijah Newren <newren@gmail.com>
>>          Signed-off-by: Junio C Hamano <gitster@pobox.com>
>>     -    Message-Id: <20180713163331.22446-2-newren@gmail.com>
>>
>>      diff --git a/t/t1015-read-index-unmerged.sh b/t/t1015-read-index-unmerged.sh
>>      new file mode 100755
>> 2:  e105e8bfbd ! 2:  d3b8d7edb6 read-cache: fix directory/file conflict handling in read_index_unmerged()
>>     @@ -59,7 +59,6 @@
>>
>>          Signed-off-by: Elijah Newren <newren@gmail.com>
>>          Signed-off-by: Junio C Hamano <gitster@pobox.com>
>>     -    Message-Id: <20180713163331.22446-3-newren@gmail.com>
>>
> <snip>
>
> Maybe this is expected or wanted (tbdiff also shows the git notes for
> what it's worth), but it seemed somewhat surprising to me.  I'd rather
> not see such "differences" displayed for the patch series that I'm
> submitting, but perhaps others see it differently?

I was surprised to see that, too, but if you have it configured I
would understand the behaviour.  

Of course, the value of showing the Message-Id: in this particular
case is dubious (by definition you won't have them on the side that
you have not posted), but if you are comparing two iterations that
have been queued (e.g. by keeping the old tips of the topic branch
unpruned by holding onto historical 'origin/pu' using reflog), it
may become useful when doing a "Wait, is that really my bogosity?
Let's go see the list archive using the message ID---ah, something
was suggested in the review and the maintainer attempted to pick it
up and squash it in, which he botched".