[openbox] Smart Window Placement?

[George De Bruin]

Dana Jansens wrote:
> I'm not sure what you're asking for, a formal explanation of the
> placement algorithm?
>   
Well, an explanation for one, and whether this is a bug or not for another.

> It doesn't top to bottom or left to right.  It collects rectangles of
> space on the screen and sorts them by size (I believe by their area),
> and it does this individually for each monitor, choosing the monitors
> in an order which can be tweaked with the new <monitor> option in
> 3.4.6, but is based on where input focus is, and where the mouse is,
> where are windows in its group/parent/children sets.  If you want
> stuff to show up in the top left of its assigned rectangle, that is
> what the placement <center> option is. Have you looked at the
> <placement> section of the default rc.xml at all?
>   
I've looked at both the default rc.xml from my package, which has:

  <placement>
    <policy>Smart</policy>
    <!-- 'Smart' or 'UnderMouse' -->
    <center>no</center>
    <!-- whether to place windows in the center of the free area found or
       the top left corner -->
  </placement>

Which didn't make me think that the placement would be at all random.  
And I looked at the website documentation (from: 
http://icculus.org/openbox/index.php/Configuration):


    Placement

<placement>
  <policy>Smart</policy>
</placement>

*policy* can be either /Smart/ or /UnderMouse/.

    * /Smart/ will cause new windows to be placed automatically by Openbox.
    * /UnderMouse/ makes new windows to be placed under the mouse cursor.


> And when something doesn't fit in any of the rectangles, it is placed completely randomly.
>   
Boy, I thought by "smart" you meant in the adjective form: witty, 
clever, intelligent.  It's obvious now that you meant it more like the 
intransitive verb form: "to cause or be the cause or seat of a sharp 
stinging pain; /also/ *:* to feel or have such a pain"

Now I understand: smart = painful = random.

> If you really want to know exactly how it works take a look at place.c
>  I've wanted to improve it with FVWM's MinOverlapPercent algorithm,
> but usually it seems just fine being random, heh, so I haven't been
> itched enough to do so.
>   
Except I don't compile from source.  Don't want to break my package 
management.  And, even if I did compile from source, it's unlikely that 
I would go digging through the code to figure out how it works.  I 
haven't been a coder for a good fifteen years now (beyond a few small 
shell scripts, and a little PHP dabbling), and haven't done any linux or 
X coding at all.

Personally, I was trying to understand this because I am writing a 
review of window managers and this behavior didn't make sense.  I mean 
personally, if I open and close a window N times, I expect the placement 
to be the same N times (assuming nothing else changes on the screen).  
But in the case of OpenBox, I was getting up to N different placements, 
and not seeing a pattern.  Personally, at first I thought I was 
imagining things...but that FireFox placement just proved that I wasn't 
imagining things.

So, I don't know about this.  It doesn't really give me a whole lot of 
confidence in being able to say that the usability of OpenBox is there.  
I can't think of a case where this could qualify as a good user 
interface design.

// George
-------------- next part --------------
A non-text attachment was scrubbed...
Name: SndChaser.vcf
Type: text/x-vcard
Size: 135 bytes
Desc: not available
URL: <http://icculus.org/pipermail/openbox/attachments/20080203/c9082e67/attachment.vcf>