[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [pkgsel] Add xresprobe, discover and laptop detect for hw detection

On Tue, Jan 17, 2006 at 12:09:55PM +0000, Daniel Stone wrote:
> On Tue, Jan 17, 2006 at 01:02:43PM +0100, Sven Luther wrote:
> > On Tue, Jan 17, 2006 at 11:55:49AM +0100, Frans Pop wrote:
> > > The description also says it only works on i386 (including AMD64 I 
> > > presume) and ppc.
> > 
> > Well, the ppc uses the OF device-tree, and if this doesn't work falls back to
> > parsing the /sys fbdev tree. This would work for all arches, but only for
> > those fbdev drivers supporting it. I know of radeonfb only right now, haven't
> > looked in a while. Also, it obviously only works in fbdev mode. Not sure about
> > vesafb, and if it provides the DDC information.
> > 
> > This could (and should if i remember the discussions of the fbdev authors
> > right) be fixed in the corresponding fbdev drivers.
> Yes, but you don't want to depend on accelerated framebuffer drivers,
> particularly if you want to use accelerated X.  In addition, most of the
> X drivers are unmaintained, let alone the framebuffers.  Let's not walk
> that path.

Well, we are speaking in the context of the graphical installer here :)

> > > It thus has a limited (though growing) application and other types of 
> > > systems will also need to be supported (probably using current manual 
> > > resolution/frequency selection methods).
> > 
> > Sure, the ideal would be for xresprobe to either work and give the right data,
> > for aall heads if posssible, or fail and inform the caller of the fail so the
> > caller can do the right thing.
> So what do you do on a laptop where you can't get any EDID for the
> external head?  What do you do in the case where you have two cards
> (let's say, one integrated and a discrete one that you actually use),
> where only one is connected?  And so on, and so forth.  I think it's
> better to get something up usefully than to insist on this ridiculous
> ideal of 'we must find out all information for every display', or
> nothing.

Well, if you can get the info, we use it, if not, well, we ask the user about
it and have some sane default in case the user has no clue.


Sven Luther

Reply to: