Re: display-buffer-reuse-frames default to t

[MON KEY <monkey@sandpframing.com>]

>>> Any objection to defaulting display-buffer-reuse-framesto t?
>>
>> Does it restrict itself to visible frames?
>
> No.  It can use iconified frames.
>
> Ideally, one would set this to 0, 'visible, or t and get the desired
> effect.  That's what I do on my branch.

Is the suggestion to extend the range of acceptable values for
`display-buffer-reuse-frames'?

IMHO there is opportunity here for some much needed changes[1].

Indeed, it would make more sense to extend the range of values allowed
to any one of: { t nil 0 visible }

I would also suggest adding `selected-frame' as an option for
consistency with `display-buffer's provision for a "specific frame" as
argument to its FRAME parameter.

Such change would require changing the custom type declaration of
`display-buffer-reuse-frames' from:

  :type 'boolean          ;; Is the declaration even correct now?

Likewise, if such change(s) are made _please_ ensure that the
variable's docstring indicate explicitly the range of possible values
and their affect on frame/buffer display, i.e. don't just punt to
`display-buffer' with a "which see".

Maybe, better is to implement an additional/alternative variable:
 `display-buffer-reuse-frames-if-graphic'

which inherits its defaults according to `display-buffer-reuse-frames'
and `pop-up-frames'?

As it is now, the current docstring of `display-buffer-reuse-frames'
is poorly specified wrt to the interaction of `display-buffer' around
`pop-up-frames' and `display-graphic-p'. Likely many users are
left with an essentially opaque customization option which often fails
to DTRT esp. where third-party authors using `display-buffer' in an
ill adapted position to appropriately accomodate user display
customizations, cf the recent report bug 7728:

 (URL `http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7728')

and discussion of `save-window-excursion' beginning around msg 68:

 (URL `http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7728#68')

[1] I use a two-head display with xrandr (and in the past with the
equivalent on w32 a three-head display) and rarely have less than two
frames open across these displays In both environments the behaviour
of `display-buffer-reuse-frames'/`display-buffer' has many weird/odd
corner cases wrt to the "strongly dedicated" and often commands do not
do what I would expect. The cummulative effect of these corner cases
is that I usually give up trying to figure out where/why/how certain
buffers/windows/frames will be displayed and just adapt to the funky.

--
/s_P\