bug#25943: 21.5 Frame Display Difficulties

[martin rudalics <rudalics@gmx.at>]

 > I have run your frame-position code a few times; I have not had time to do
 > anything else.

I suppose you did not (or were not able to) apply my patch so

(setq x-gtk-use-window-move t)

had not effect on the outcome of your runs.  Right?

 > I am unable to do anything similar for 23.2 because
 > function frame-position does not exist in 23.2.

I forgot.  Instead of

	 (pos (frame-position frame))
	 (left (car pos))
	 (top (cdr pos))

use

	 (left (frame-parameter frame 'left))
	 (top (frame-parameter frame 'top))

for Emacs 23.

 > Mostly, but not always, the 9-C (fun) frames are consistently in their
 > correct corner.  This is true for the 1-4 (arg) frames as well; I do not
 > have any record of a deviation, so arg frames may be perfect.  The 5-8
 > (par) frames are the deviants, which seem to go the the Left Top corner.

Your results for 25.1 are better than mine.  Here only the "arg" runs
position correctly.  The "fun" and "par" frames all end up in the top
left corner.  So if you do want to continue working with an unpatched
Emacs 25.1 and want more or less correct positioning you will have to
use the "arg" notation.  Otherwise you will have to either patch your
25.1 or switch to the current development version.

(
  (arg
   ...
   (name . "4 Right Bottom")
   (p-left . -40)
   (p-top . -40)
   (left . 1181)
   (top . 694)
   (width . 816)
   (height . 400)
   (r-left . -51)
   (r-top . -58))

It might be interesting to see where these differences in the "arg" case
come from - 11 pixels horizontally and 18 pixels vertically.  What does

M-: (frame-geometry)

in that frame return?

Thanks, martin