Peter Korn wrote:
> I'm cc-ing the GNOME Accessibility list, where folks who are interested
> in this live.
>
> The GNOME On-screen Keyboard is being replaced with Caribou (see
> http://live.gnome.org/Caribou). There remains a need for a way for
> alternate input devices (which electrically look like mice, and which
> present the USB mouse HID) to connect to a UNIX desktop and drive
> accessibility software like Caribou. Various workarounds (e.g. moving
> the mouse to the upper left corner - assuming the keyboard lives there)
> make this non-critical, but still important.
>
> I don't know enough about Xi2 to say whether that meets all needs or
> not. Perhaps the Caribou maintainer might comment. One concern though
> is whether Xi2 is a good path forward for thin clients like Sun Ray, or
> whether we need to use something like libusb for that.
libusb doesn't seem like a good solution to me, since it replicates a bunch
of work already done in the X server & kernel, and may have issues with device
permissions and access, as well as failing for remote access & virtualization
scenarios. I understand it was considered as a workaround for the Sun Ray
design, where the Sun Ray device firmware combines all HID devices into a single
logical device, but if you need independent logical devices controlled by
independent physical devices, I'd rather see the Sun Ray firmware design changed
than force every program that uses Xi2 to have special case code to call libusb
directly just for the Sun Ray case.
However, my understanding of the typical onscreen keyboard usage model (from
what I remember of the interactions years ago with the gok team), is that you
normally need the physical HID pointer device to control the onscreen keyboard
client, and it in turn controls the core X protocol pointer that all the other
clients interact with. In the Xi2 world, I believe this is simply a case of
detaching the physical HID pointer device from the Virtual Core Pointer device
and having it be a separate Xi2 extension device that the onscreen keyboard gets
input from via Xi2 and then controls the VCP to interact with other clients.
(And thanks to the MPX facilities in Xi2, you should be able to see two cursor
images on screen for that, each with it's own pointer focus.)
More information on Xi2 can be read in the creator's blog at:
http://who-t.blogspot.com/search/label/xi2
--
-Alan Coopersmith- [email protected]
Oracle Solaris Platform Engineering: X Window System
_______________________________________________
xwin-discuss mailing list
[email protected]
List info: http://mail.opensolaris.org/mailman/listinfo/xwin-discuss
Unsubscribe: http://mail.opensolaris.org/mailman/options/xwin-discuss
