bug#46990: 28.0.50; popup menu not navigable via arrow keys on lucid build

[Colin Baxter <m43cap@yandex.com>]

>>>>> Eli Zaretskii <eliz@gnu.org> writes:

    >> From: Colin Baxter <m43cap@yandex.com> Cc: contovob@tcd.ie, ,
    >> 46990@debbugs.gnu.org Date: Wed, 10 Mar 2021 19:39:23 +0000
    >> 
    >> > Can you or someone who sees the problem please bisect this?
    >> I'm > delaying the release of Emacs 27.2 in the hope that we
    >> could fix > it, if indeed this was introduced between 27.1.90 and
    >> 27.1.91.
    >> 
    >> I may be guilty of crying wolf when there is none. If I configure
    >> the 27.1.91 pretest for lucid and "make", the effect does occur
    >> on launching emacs via src/emacs. However if I go the next step
    >> and "make install", the effect does not occur when I launch the
    >> installed emacs. I will double-check this by installing from the
    >> git master branch and report back.
    >> 
    >> I am sorry. I stupidly assumed launching emacs from an src
    >> directory would give the same beast as launching from the
    >> installed bin directory.

I can now confirm that if I install the git master branch and launch
emacs from the installed bin directory, I can navigate the lucid
menu-bar via the arrow keys and the mouse. However if I launch emacs from
the git/src directory then I can not navigate the menu-bar via the
arrow keys. I can only navigate via the mouse.

It also explains why I thought the effect only occurred in the second
pre-test and not the first. I remember that I installed the first but
not the second.

    > No need to be sorry: your assumption was correct.  Emacs should
    > generally work the same when you run it uninstalled.  It is
    > strange that it doesn't in this case.

    > It is true that the fact the problem doesn't happen in the
    > installed Emacs makes the problem less serious.  But since we
    > don't really understand why it happens in the uninstalled Emacs,
    > we cannot guarantee that it will never happen in an installed one,
    > either.

It could an artefact of my system: 4.9.0-14-686-pae #1 SMP Debian
4.9.246-2 (2020-12-17) i686 GNU/Linux. Perhaps someone needs to confirm
the finding on a different system.

    > So if you can afford bisecting this, or debugging it
    > further, I'd appreciate any additional information you could dig
    > out.  Ideally, we should at least understand why this happens,
    > before we decide whether it should delay the release.

I'm not confident I can bisect, having never done it. It depends on the
available times, though, but certainly I will try.

Best wishes,