bug#20552: 24.4; cc

[gottlieb]

On Thu, May 14 2015, martin rudalics wrote:

>> As you wrote it there is no difference.  (left . 0) remains
>> about 1/8 inch right of flush left (even if executed twice).  BUT
>>
>> (modify-frame-parameters nil '((user-position . t) (width . 176)
>>                                 (left . -1300))
>>
>> does become flush left on an initial emacs -Q whereas without the
>> user-position (which I will now read about) it required two executions
>> to become flush left.  Recall, previously the first execution left the frame
>> several inches right of flush left but a second execution made it flush
>> left.
>
> Still weird.  What happens with
>
> (modify-frame-parameters nil '((user-position . t) (width . 176) (left . 1)))

I am not at the 2560x1600 screen.  But I tried (left . 1) anyway.  It is
still about 1/8 inch from flush-left.  I tried (on this 1680x1050
monitor) (left -1300) and tried (left . -500) both gave flush left.

Conclusions
1. (user-position . t) is needed to get flush left immediately.
   The manual is clear on this if you know to look at user-position.

2. (left . 0) is not quite flush left even with user-position.
   Same for (left . 1).  Perhaps the window manager is causing this.

3. (The weird part) without user-position (left -1300) is not close to
   flush left when executed once but is flush left if repeated.  Its as
   though the second time it carries the force of user-position despite
   user-position not having ever been specified on this fresh emacs -Q.

> Also, you could try playing with the `(+ POS)' and `(- POS)' position
> specifications (see section 28.3.3.2 Position Parameters of the Elisp
> manual).
>
> martin

I tried those before I knew about (i.e., you told me about)
user-position and I still could not get flush left.  I will try again
with (user-position . t).

Thanks again,
allan