bug#10417: 24.0.92; Shell completion regression for shell-completion-execonly

[Lars Ingebrigtsen]

Chong Yidong <cyd@gnu.org> writes:

> Suppose I have a non-executable file named screen.el in the current
> directory.  With latest trunk:
>
> emacs -q
> M-: (setq shell-completion-execonly nil) RET
> M-x shell RET
> ./scr
> TAB
>
> Emacs says "No match" in the minibuffer.  With Emacs 23, the TAB
> completes "./scr" to "./screen.el", which is the expected result for
> shell-completion-execonly nil.

This was a decade ago, and I'm unable to reproduce this in Emacs 25.1
(or the current trunk).

But this bug report was reopened (by Eli) as a response to bug#42383, so
... perhaps I'm testing this wrong?  Or has this been re-fixed since it
was reopened a year ago?  (There has been changes in
shell-command-completion-function the last year, but I'm not sure
whether that explains it.)

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no