steven@uicsgva.UUCP (03/15/88)
I'm curious: we're running Ultrix 2.2 on GPX/II's. The 11.2 distribution says that you can build the distribution for Ultrix 2.0 systems. But 2.2 systems don't seem to be easily modified to use the stuff in ./server/ddx/dec/qdss. Specifically, DEC has a completely different setup for the kernal than 4.3. You can't replace init.o with the one in the server directory because of a plethora of rearrangements. Does this mean we are stuck with their version (10.4), or can we omit applying these changes to the kernal? Is the new code faster? Steven Parkes Coordinated Science Lab University of Illinois steven@csg.uiuc.edu
jg@jumbo.dec.com (Jim Gettys) (03/17/88)
At this particular instant, I am running exactly the R2 release on my GPX (actually a 3200, but I had it running on a uVaxII last week before the 3200 showed up) on Ultrix 2.2. You have to do precisely NOTHING to the base X distribution or to your kernel. I had personally installed a virgin 2.2 on the machine before installing the MIT release, and not installed any V10 X version at all. Just follow the installation directions; don't go out of your way looking for trouble that doesn't exist. The build of the core distribution went without error, and installed without error (make sure that you have the ULTDCMT022 subset installed; the X distribution uses soelim for the man pages), and runs. Do not fuss with your kernel, though once converted to V11, you can remove the Xos stuff from your kernel, if you had been running that. Jim Gettys
reilly@atari.DEC.COM (Michael Reilly) (03/21/88)
No changes are need to the kernel for Ultrix-32 2.2 to run X11R2 on a GPX. The
only change needed is to use the qd.o driver (found in /sys/BINARY.vax)
from V2.0 of Ultrix-32 on V2.2.
To run X11R2 on a GPX -
Make a backup copy of the file /sys/BINARY.vax/qd.o
copy the file /sys/BINARY.vax/qd.o from the V2.0 system into the
directory /sys/BINARY.vax on the V2.2 system.
Edit the configuration file for your system in /sys/conf on the
V2.2 system to remove the references to {xos, XOS}
Save a copy of the running kernel. I mv'ed mine to /vmunix.x10
Make config, etc. to rebuild the kernel in the normal manner.
Boot the new kernel and bring up X.
The above steps took me 30 minutes (I was careful to save all of the
original files in case I messed something up along the way. Of course
building and installing X took a little longer (6 hours).
This is the method I use to run X11R2 on my GPX. Digital Equipemnt
Corporation is not responsible for this posting nor does Digital endorse
this procedure.
mike
Michael Reilly @ Digital Equipment Corporation
Networks and Communications Group
reilly@atari.dec.com (or reilly%atari.dec.com@decwrl.dec.com)
...{allegra,decvax,ucbvax}!decwrl!atari.dec.com!reillyguido@cwi.nl (Guido van Rossum) (04/04/88)
I have the same experience (very smooth installation) on a VaxStation 2000 with X 11.2 Ultrix 2.0, except for one thing: apparently a bug in the optimizer breaks xterm and xbiff. The remedy: turn off -O. Since Jim Gettys reports no problems on Ultrix 2.2, DEC seems to have fixed this bug since, so I won't go out of my mind to find more details. --Guido van Rossum, CWI, Amsterdam -- guido@cwi.nl
kwh@sei.cmu.edu (Kurt Hoyt) (04/05/88)
In article <464@sering.cwi.nl> guido@cwi.nl (Guido van Rossum) writes: >I have the same experience (very smooth installation) on a VaxStation >2000 with X 11.2 Ultrix 2.0, except for one thing: apparently a bug in >the optimizer breaks xterm and xbiff. The remedy: turn off -O. Since >Jim Gettys reports no problems on Ultrix 2.2, DEC seems to have fixed >this bug since, so I won't go out of my mind to find more details. > >--Guido van Rossum, CWI, Amsterdam -- guido@cwi.nl That's interesting. I have installed X11.2 on my uVAX II under Ultrix 2.0 with -O turned on for all the compilations and I have not experience the xterm or xbiff problems. So, if our optimizers are the same, your problem is elsewhere. I hope to go to 2.2 soon -- and I will be recompiling everything, even though the 2.0-generated binaries run fine on our 2.2 systems. -- Kurt Hoyt, Software Engineering Institute -- kwh@sei.cmu.edu "From previous research with rendering systems we have learned that a good dose of gratuitous partial differential equations is needed to meet the paper quota for impressive formulas." SIGGRAPH '87 Proceedings, p. 73