Re: _NET_WORKAREA and dual head

[Billy Biggs <vektor dumbterm net>]

Rob Adams (readams readams net):

> On Fri, 2004-03-12 at 10:59 -0600, Billy Biggs wrote:
> 
> > Just looking around at the dual head setups here, people seem to
> > like panels on the inner edges of the monitors since it's quick to
> > access from both displays.
> 
> Really?  I've never seen anybody do this.  This setup is not
> well-supported right now because of the way the struts work.

  It seems well liked and works with Windows XP.

> >  You can determine this simply using the XINERAMA information and
> >  the strut rectangles.
> 
> Actually, you can't.  The way the specification is defined the struts
> are specified in root window coordinates, and are not limited to an
> xinerama. [...]
> 
> As current written, the specification does not allow internal struts
> with anything like reasonable semantics.
> 
> We have the _STRUT_PARTIAL property that have sufficient flexibility
> to enable full-fledged struts on non-internal xinerama edges, though
> it requires a fair bit of algorithmic sophistication to implement this
> in the window manager and is a bit of a pain to do correctly for a
> client as well.  At the time, however, we didn't deem internal edge
> struts sufficiently important.

  I'm sorry, I thought that _NET_WM_STRUT was deprecated in favour of
_NET_WM_STRUT_PARTIAL.  Furthermore, I thought metacity at least by
looking at it has these get_work_area_xinerama calls which on first pass
seemed to be built to handle these cases.  It shouldn't be any more
difficult than maximize logic with a panel on just one head.

  -Billy