home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!sdd.hp.com!think.com!ames!ig!pcs.cnu.edu!shendrix
- From: shendrix@PCS.CNU.EDU ("", Shannon Hendrix)
- Newsgroups: comp.os.coherent
- Subject: Re: Tcl to replacement most of /bin & /usr/bin (was: Tcl on Linux
- Message-ID: <9211202320.AA01986@PCS.CNU.EDU>
- Date: 20 Nov 92 23:21:08 GMT
- References: <9211201646.AB00614@PCS.CNU.EDU>
- Sender: daemon@presto.ig.com
- Reply-To: "Coherent operating system" <COHERENT@indycms.bitnet>
- Lines: 34
-
-
- Building the graphics engine into the kernel doesn't require the
- details of each card to be included any more than it does if the engine
- is separate like X. The only difference is that it is now a kernel
- process that always gets the horsepower it needs. I/O and system tasks
- are atomic and run to completion (though disk I/O can be preempted for
- obvious reasons...). The basic UNIX scheduler is not designed to handle
- something like a graphics server. What I am talking about would still
- use graphics drivers but would not run as a user process.
-
- If I'm not mistaken, the approach I'm talking about is being
- used for PowerPC, some versions of Mach, the Apple Macintosh, and
- Windows NT. OS/2 2.0 has hooks in its OS for the graphics engine. If
- you design a kernel with the graphics system in mind its going to be
- better. The UNIX kernel was not so until it gets changed it won't run
- graphics as well as it could.
-
- It occurs to me that you could still make X a user process if you
- just had graphics primitives built in to the kernel. That way you wouldn't
- be stuck with X. This would also make it work on differing architectures.
- While my PC doesn't have a frame buffer mapped to memory like a Sparc or
- something (or mapped to a /dev/ entry) it is possible. GNU C maps the
- graphics card buffer as a single 1 meg address space in upper memory
- allowing you to access it just like a Mac or a UNIX box. No more stupid
- segment swapping. This might help promote better motherboards too. I'd
- like to see a 386/486 class motherboard that was designed to be a good
- motherboard and not a good DOS motherboard. That might take awhile but
- maybe DOS will die someday...
-
- =====================================================
- Shannon Hendrix |shendrix@pcs.cnu.edu
- Christopher Newport University |---------------------
- Newport News, VA |** space for rent **
- =====================================================
-