<aaronp> Sorry I've been behind on this stuff.  I successfully built your prime branches and our driver against the SDK bits and can successfully start an X server with xf86-video-intel or our driver on my Optimus test board.  I haven't gotten around to wiring up the provider stuff yet, though.
<aaronp> In terms of breaking our driver horribly, you fail at that.  :P
<keithp> aaronp: we'll have to try harder, I guess?
* ssp_ (ssp@nat/redhat/x-wlesbjbhtarhvpfr) has joined #xorg-devel
<aaronp> How do you plan to deal with different GLX visuals / fbconfigs between providers?
<mlankhorst> aaronp: so does nvidia do intel y swizzling or not
<keithp> aaronp: as airlied described, applications that want to play must deal with the GLX extension that tells them to essentially reload GL
<keithp> if you've got non-participating applications running, you can't switch
<keithp> so, it appears to punt to the application
<aaronp> mlankhorst: I haven't managed to track down an answer to that.
<keithp> which seems sensible
<mlankhorst> aw
<aaronp> keithp: so apps have to re-query the fbconfig list when their contexts go away?
<keithp> I'd assume so; how else could it work?
<aaronp> Okay
* keithp hasn't read that GLX/EGL extension to know though
<aaronp> That seems like it should work.
<aaronp> Mucking about with the Visuals might be harder, but hopefully doable.
<keithp> yeah, essentially re-do all of their GL configuration
<mlankhorst> keithp: iirc nvidia wrote it
<keithp> oh, I don't know about the visuals; doesn't seem like airlied cares
<keithp> so, use fbconfigs or fail
<keithp> (the whole visuals hack is ugly anyways)
<aaronp> Yeah, it's pretty gross, but workstation apps need it.
<aaronp> (workstation apps need a lot of hand-holding)
<keithp> they probably won't be playing