nanogui: Suggestion for adding nanowm into graphic server!
Subject:
Re: [nanogui] Suggestion for adding nanowm into graphic server!
From:
Alex Holden ####@####.####
Date:
5 Aug 2002 15:21:35 -0000
Message-Id: <3D4E94B6.7060203@linuxhacker.org>
Greg Haerr wrote:
> remain exactly as it is now. Instead,
> a small internal WM* api could be created,
> which uses a callout mechanism to send
> events to a linked-in portion of the program.
> Comments?
I was just wondering, how about instead of implementing a specific WM
API, instead allow arbitrary multiple application threads to run
cooperatively within the server, optionally at the same time as having
remote applications (not using OS threads, but instead using
GrGetNextEvent() as the context switch point). This would allow some
number of applications linked into the server to run simultaneously with
some number of remote applications. There would be a compile time option
to completely remove the networking support like now, but
LINK_APP_INTO_SERVER mode wouldn't be limited to a single application
any more.
The biggest stumbling block I can see is how to allow multiple
applications to be linked in and initialised, without resorting to
dynamic loading. Perhaps each app which gets linked in would need to
have its own global, uniquely named version of main(), and the real
main() would call all of the client main() functions from a
user-configured start_clients() function it calls after the main server
initialisation and before starting to accept remote clients.
Just an idea.
--
------------ Alex Holden - http://www.linuxhacker.org ------------
If it doesn't work, you're not hitting it with a big enough hammer
