Re: master 18b680cfd1: Fix bug#52467 by adding a new custom variable 'display-comint-buffer-action'

[martin rudalics]

> Some commands pop their windows in unexpected places anyway,
> so often have to resort to typing a same-window-prefix key
> before running such commands :-)
>
>> The attached diff should illustrate what I mean here.
>> [...]
>> +(defvar display-buffer--reuse-or-same-window-action
>> +  '((display-buffer-reuse-window
>> +     display-buffer-same-window)
>> +    (inhibit-same-window . nil))
>
> Shouldn't the first item be display-buffer--maybe-same-window
> before display-buffer-reuse-window?  This is what is used by default
> in display-buffer-fallback-action, but it has a comment with a question,
> so probably it's really redundant:
>
>    (defconst display-buffer-fallback-action
>      '((display-buffer--maybe-same-window  ;FIXME: why isn't this redundant?
>         display-buffer-reuse-window

'display-buffer--maybe-same-window' is here for compatibility reasons
with the 'same-window-buffer-names/-regexps' options and can be used
only in an "ored" fashion as in 'display-buffer--maybe-at-bottom' or
'display-buffer-fallback-action'.  Otherwise, it would defeat
'display-buffer-alist' because it uses the same window only if that has
been advocated by 'same-window-p' which ignores 'display-buffer-alist'.
It's not redundant as long as we want to maintain compatibility for
those options.

> Grep counted 100 occurrences of pop-to-buffer-same-window in the source tree,
> but I guess only shell commands could use pop-to-buffer-reuse-or-same-window?

It can be used as a substitute for 'pop-to-buffer' wherever we decide
that consistency should prevail over long-standing behavior.  I have
been missing the present and the preceding thread but I think we should
revert the behavior for Emacs 28 and use 'pop-to-buffer' there.  Sam's
protest in this respect was justified - we cannot simply tell people
that they should customize 'display-buffer-alist' in order to get back
some long-standing behavior.  That's not our style.

Now 'pop-to-buffer-reuse-or-same-window' is still no 100% substitute for
'pop-to-buffer' but we can try it because it handles Sam's main argument
against 'pop-to-buffer-same-window' - that the same buffer appears twice
on the same frame.  Whether that suffices to fix all regressions in this
context should probably be answered in Emacs 29.

martin