Subject: Re: XF86Config for Sparcstation 20?
To: None <port-sparc@netbsd.org>
From: der Mouse <mouse@Rodents.Montreal.QC.CA>
List: port-sparc
Date: 11/20/2004 12:49:04
>> [I]f the framebuffer can't do [RENDER] in hardware or firmware, it
>> makes little odds whether the software rendering is done in a server
>> software RENDER implementation or elsewhere.
> If it is done on client side, there are much more data flowing
> between the server and client.  [...slow network...]

True enough.

>>> Also, how do I exit the Xserver, if it is not run from xinit?
>> You normally don't.  Aside from deliberate by-hand runs, [...]
> What's wrong with by-hand runs?

Nothing at all.  They're just rare.

> This is good for testing, and also for converting a slow Sparc box (a
> 20 MHz Sparcstation SLC in my case) to a X terminal: X -query <xdm
> server> .

Yes.  That's what -once is for.

>> The server normally runs until it resets upon losing its last
>> client, or (if using xinit) until xinit sees the client script exit
>> and kills the server.
> How to make the client script exit if gnome-session, kde or some
> other desktop monster just hangs?

That's one reason I would recommend against making anything you use a
lot (such as your window manager) your distinguished client.

But in the face of that kind of bug, I see nothing wrong with dropping
to ddb and killing something (I usually go for the server).

> But, if the Xserver is running, there is no way to use the console,
> even if there is no difference between text mode and graphics mode.

This is not quite true.  The only reason the console is inaccessible is
that the keyboard has been stolen away from it.  If you have something
like my kmmux layer in between the hardware and the X server, you can
switch the real keybaord to a different virtual keybaord and use the
console normally.  Also, if you have a second keybaord on the machine
(unusual but not impossible), normally X will claim only one keyboard.

> What I'm generally trying to say here is that there are many often
> useful features in XFree86 (and many other will probably come with
> the new X.org Xserver) which are not related to x86 hardware,

Agreed.

> so wanting to have XFree86 on non-x86 platforms is quite reasonable.

Except that I think you don't so much want XFree86 as want those
features.

It may be that trying to bludgeon XFree86 into running is the cheapest
(and I don't mean money) way of getting them.  But it also may not.
For example, it would be fairly easy to diddle the keyboard handling
code to put a nerve-pinch akin to ctl-alt-backspace into the x.org
server.

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B