xcompmgr -- Proposal 2: ARGB-window dropoff shadow

[Yann Droneaud]

(First, i have to apologize: in an earlier mail, I showed screenshot
made with server side shadows (xcompmgr -s), instead of client side
shadows (xcompmgr -c)).

Le vendredi 25 septembre 2009 à 19:31 +0200, Eeri Kask a écrit :

> As it looks like there really seem no apparent _technical_
> restrictions involved to not draw a drop-off shadow around
> transparent or heavily XShaped client windows?
> 

Indeed, as you said, there's no technical raison, but it's ugly.

See those new screenshots:

using xcompmgr -c:
https://quoi.quest-ce.net/data/xcompmgr/xcompmgr-c.jpeg

using xcompmgr -cA (your patch applied, the shadow is here):
https://quoi.quest-ce.net/data/xcompmgr/xcompmgr-cA.jpeg

same content, but borderless window:
https://quoi.quest-ce.net/data/xcompmgr/xcompmgr-cA-borderless.jpeg

more: a KDE tooltip with round corner (don't ask me why the window is
larger than the content, it's should not be needed, it's only waste of
space, thanks KDE):
https://quoi.quest-ce.net/data/xcompmgr/xcompmgr-cA-tooltip.jpeg

So my point is still valid: shadows are bad when they are applied around
the window (rectangular) and not the content.

IMHO, the server side shadow is a better implementation.
https://quoi.quest-ce.net/data/xcompmgr/xcompmgr-s.jpeg


> Then, how to proceed in the case of "conflict of interests" if, as
> opposed to some particular client window, the computer user still
> prefers a shadow even around semitransparent or XShaped windows
> (i.e. beyond the rectangular area (in the sense of the "X-window
> attribute") enclosing the window and its border)?  :-)
> 

A property would do the job. But it's a bit late to standardize such.

> Apart from the fact that semitransparent (and XShaped) windows in
> general need not have a nonrectangular outer shape at all,
> therefore as long as there is no property-based mechanism for
> client-based shadow on-off configuration yet, and in fact even
> unrelated to that, it looks quite reasonable to let the computer
> user resolve this above conflict (by some xyz command line
> parameter), don't you agree?  :-)
> 

I'm agree.

> (Of course let's keep xcompmgr's default behaviour to not draw
> shadows around ARGB windows but introduce a method to override this
> default.)
> 

I'm OK with this patch.
Sorry for the comments, but it was an opportunity to explain my taste
about shadows.

Regards.

-- 
Yann Droneaud