bug#72059: [PATCH] Ensure that git diffs without signature (--) are properly identified

["Kévin Le Gouguec" <kevin.legouguec@gmail.com>]

Luis Henriques <henrix@camandro.org> writes:

>> Also, how does this work for messages containing multiple patches? Is
>> detection of just the start of each patch enough?
>
> Do you have an example where this happens?  I don't think I ever saw an
> email with two inlined patches.  But obviously, with this patch applied,
> everything from the "^diff --git " up to the end of the email will be a
> diff.  Just like everything after "^=== modified file " or "^Index: " will
> be a diff.

FWIW, I've had this uncommitted kludge in my config for some time now:

  @@ -403,6 +431,9 @@

   ;; TODO: handle inline diffs in Gnus articles; maybe tweaking
   ;; mm-uu-type-alist?
  +(with-eval-after-load 'mm-uu
  +  (setf (nth 1 (alist-get 'git-format-patch mm-uu-type-alist))
  +        "^-- \\|^$"))

Past Me wasn't considerate enough to add a Message-ID demonstrating
where this fixes problems.  An earlier version of that config featured:

  ;; Gerrit does not include trailing "-- ".
  (with-eval-after-load 'mm-uu
    (setf (nth 1 (alist-get 'git-format-patch mm-uu-type-alist))
          "^-- \\|\\'"))

Past Me eventually ditched this for being, citing the commit message,
"too flaky".  Again, did not bother to add a reference to a problematic
sample 🙄

I think the logic behind "^-- \\|^$" is that ^$ ought to find the end of
a "very-inlined patch" (i.e. not "attached", not even as an inlined
part; think 'C-x v d', 'C-x h' 'M-w', 'C-x o', 'C-y') because even empty
context lines have a leading space…

… but that's just guesswork.  Thought it'd be worth mentioning, but feel
free to ignore the noise if not helpful.