Re: [gdm-list] GDM - gdmdynamic and Sun Ray support
- From: Jerry Wall <Jerry Wall Sun COM>
- To: William Jon McCann <mccann jhu edu>
- Cc: gdm-list gnome org, Bob Doolittle <Robert Doolittle Sun COM>
- Subject: Re: [gdm-list] GDM - gdmdynamic and Sun Ray support
- Date: Thu, 11 Oct 2007 08:51:42 -0700
William Jon McCann wrote On 10/11/07 08:31,:
> Hi Jerry,
>
> On 10/11/07, Jerry Wall <Jerry Wall sun com> wrote:
>
>>The issue with Sun Ray and other thin clients is that there
>>is no local hardware and no kernel drivers involved in
>>plumbing one into the system. They are strictly network
>>connected client so there is nothing to tell something
>>like ConsoleKit about.
>
>
> Interesting. The SRSS 4.0 Administrators guide (Chapter 4) seems to
> indicate otherwise. What would be really useful then is a description
> of exactly how Sun Ray client and server software works and a link to
> the code. Would you mind providing them?
>
> It is pretty important that we understand what the particular needs are.
>
Sorry but we cannot provide a link to the code at this time.
The Sun Ray uses a set of network protocols to communicate
with servers that have the Sun Ray server software loaded
on them. The server software provides the glue into the
system to authenticate a Sun Ray client, attach it to
a user's desktop session and communicate with local
USB devices plugged into the Sun Ray. The base
functionality of the client can be (and is) used
without having any Sun Ray specific kernel
modules involved. All of the mapping and communication
is done in user land via normal TCP/UDP socket
operations. One can have an X based desktop
session without any Sun Ray kernel modules.
To provide some legacy device emulation
we do deliver pseudo drivers that allow serial,
parallel, mass storage and audio applications to
think there is a local hardware device but these
pseudo devices are very ephemeral and do not
persist over the life of a user desktop session.
There are no kernel device associations at the start
of a desktop session and often there are no
current kernel device associations during the session.
In general there are enough issues with creating
pseudo drivers and devices that we try to avoid
it whenever possible. That is why we push libusb
as the interface for applications wanting to access
the Sun Ray USB devices since our libusb implementation
talks TCP/IP directly to the Sun Ray client without
using any USB kernel drivers. Even the current
pseudo drivers are canidates for removal, our
thinking about how to support ALSA on linux is to
remove the current OSS pseudo driver and to deliver
a library module that connects the ALSA framework to
the audio portion of the Sun Ray protocol.
--Jerry Wall
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]