On Tue, Aug 31, 2004 at 02:32:49PM -0500, Branden Robinson wrote:
> retitle 253607 xserver-xfree86: [debconf] don't set monitor sync ranges in 
> config file at all; just rely on DDC/EDID to work
> severity 253607 wishlist
> thanks
> 
> > Like said, since XFree 4.3.0, Xfree, at least on the graphic drivers
> > that support it, is capable to run without monitor timings, and able to
> > ask the monitor this over the ddc protocol.
> 
> ...except you don't always get an answer back.

Sure, but the choice of setting a empty monitor timing line should be
possible, and it is not currently.

> > Ideally, the configuration stuff should somehow interogate the graphic
> > card/monitor, and see if it is able to use the ddc protocol to get the
> > datas, tell the user that this values were detected, and ask if he want
> > to use the default values, autodetect each time, or provide some values
> > himself, the default being autodetecting stuff.
> 
> Yes, "somehow".  You don't provide any code, or anything beyond the barest
> sketch of a design really, so this is going to have to wait until after
> sarge.

Fine with me. It would be nice if at least for sarge it would be possible to
set an empty monitor timing entry.

> > By letting these entries absent of the config file, at worst X will
> > revert to some VGA standard mode, so there is no danger, but you may get
> > reports about "why are the icons so big" from people without monitor
> > attached, or using broken monitor switches or such.
> 
> VGA standard mode is far too crude for most purposes today.  Modern
> desktops expect to have a ton of real estate, which is why I have added
> 1152x768 and 1024x768 to the fallback default mode list in the current
> "debconf overhaul" work being done in XSF SVN.

Yeah sure, VGA standard mode is too crude, but it is better than not having
any output at all, which is what i get when i accept the default choice on my
LCD panel.

> > I guess read-edid can be used for just this, but then maybe it is x86
> > only, is that possible ?
> 
> Actually, Colin Watson has done some work to make read-edid operate on
> PowerPC as well[1].  I hope to get that patch applied in time for sarge; I
> have authorized an NMU for someone else to do so if people don't think I
> will get to it in time.

Cool.

> On a more general note, now that the licensing of the X-Oz autoconfig stuff
> has been clarified (it did exist under traditional MIT/X11 terms when it
> was first checked into XFree86 CVS), it may be that in sarge + 1 when we
> migrate to X.Org we will use "X -autoconfig" instead of (or as an
> alternative to) the combination of discover, mdetect, and read-edid.

Yeah, nice. I had some hope with the configuration-file-less solution David
was working on, but with the licence mess, i guess that is lost to the free
world forever, oh well ...

> My only concern about that approach is we'd better be damn sure starting X
> doesn't lock up people's machines, especially on our more exotic
> architectures.  discover and mdetect don't, and read-edid only seems to
> hurt x86 users with buggy hardware.

He. That said, there is one problem with the read-edid thingy, namely when you
use a monitor switcher which is not relaying the DDC lines, or you start X
while the monitor is off -> you are screwed.

Thanks for taking care of my bug report.

Friendly,

Sven Luther


Reply via email to