danny@itm.UUCP (Danny) (04/02/85)
X
Well, I brought up XELOS for the first time this weekend. It didn't
take me too long, about 2 hours. The documentation was well written,
if not entirely thorough, nor correct.
I followed the directions in the "setup" section of the Adminstrator
Guide, and tried to bring XELOS up multi-user, run level 2. Run
levels in general are a new concept to me, and weren't well explained.
Apparently, you can run any level you wish, 0-6, but 2 is default for
general-purpose multi-. Possibly, another level might be used during
backup, to ensure file-systems are quiet, etc. But, I digress.
Xelos became very confused when trying to run at level 2. Cron started
and immediatly aborted, fsck found garbage in file sys /u, etc.
I booted it into single user again, and started poking around.
According to the documentation, the 4th file of the first tape is
the /usr distribution in cpio format. In fact, it is the root
file sys in cpio format, and the entire second tape is /usr!
Thus, /usr/lib/crontab didn't exist! It *looked* fine, because
there *was* a /usr/bin, and a /usr/lib. So be forwarned! Our
distribution was on 2 tapes, at 800 bpi. So much for the new
documentation being accurate!
As for the incompleteness, the file /etc/checklist tells fsck(1M)
which filesystems to check. One entry is for /dev/dsk/0s2, where
/u will reside (by default). The setup section gave no mention
of how to go about setting up /u, what its size may be, etc.
Only after more digging did I find the disk layout for d256's (dsk(7M)).
By subtracing the beginning slice block numbers, I determined that
sizeof (/u) == sizeof (/usr). From that, mkfs(1M), labelit (1M), and
mount(1M) all worked, and fsck no longer complained.
This time, bringing the system up to run level 2 went flawlessly.
The system seems fast (what do I know? the 3210 makes a pretty
nice PC!). Cc(1) is much slower, however. The obligatory
"Hello World!\n" program took 19 seconds (real) time to compile.
Under edition7, the same program takes 9 seconds. In talking to
one guy at the P-E booth at UniForum, Dallas, he mentioned that
the slowness is due to the flexnames (any length name may
be used, but only the first 100 chars are significant!).
More is no more. But, pg does everything more did, and (dare I
say it?) more. It can peruse backwards, search for regular expressions,
go to any page, display a user-defined prompt, and clear the terminal for
each new page.
Another plus was that in one of the messages XELOS gives on comming
up is the amount of available user memory. We will have about 100K more
than under edition7. I assume that the ripping of the multiplexed file
code contributed greatly to the smaller size.
I did not try to read in any tar tapes, nor any restores from
edition7. I'll save that for another day.
All in all, I'm excited about all of this. The documentation
is very readable, and explains most things very well. Of course
the best written documentation is worthless if it's wrong.
If the software does *half* of what it claims to, it will be
a real joy to use!
I'll post more info as I experience it. I don't anticipate
any problems, but we'll see!
Danny
--
Daniel S. Cox
({gatech|akgua}!itm!danny)