SCO flames (and V.4 comparisons)

Sean Eric Fagan sef at kithrup.COM
Mon May 6 19:05:18 AEST 1991


In article <1991May6.002054.15900 at ico.isc.com> rcd at ico.isc.com (Dick Dunn) writes:
>Uh-huh...isn't this the same Sean who was raising hell about a week ago
>because people were flaming SCO? 

Yep.  And to a certain select portion of this readership, I would like to
take the oppurtunity to apologise.  I could claim I was tired (since I was),
but the truth is, I would act the same way during almost any time.  A
knee-jerk reaction, compounded by two years' worth of keeping my trap shut
8-).

>So apparently only SCO's engineers have feelings--is that it, Sean?  I
>don't like the flames that have been directed at SCO, but don't expect
>any sympathy until you show as much civility as you ask.

The only people I'm angry with in regards to SysVr4 are those who decided
that everything should go into the kernel.  At one point, AT&T *boasted*
that SysVr4 contained over a million lines of C! *shudder*  Talk about
maintainance nightmares.  From what I've been able to tell, the engineers at
Dell did a rather good job with SysVr4, and I applaud them for it.  I don't
know what ISC did with it, because I haven't really heard about it yet, I
don't believe.  I'm sure they did a fine job, too (although there will, of
course, be bugs, just like every other system, and I have not yet seen a bug
in a shipped system that I think calls for insulting the company, not even
the security bug in 3.2).

Now, for a bit of explanation about my problems with SysVr4.  First of all,
I am *positive* that, with a bit of tuning and work, SysVr4/386 can be made
to be at least as fast as 3.2v2 or whatever ISC's equivalent product is
(sorry; I don't keep track 8-).  And I'm about to fall over asleep 8-;); it
can certainly be made more usable that stock 3.2v2, since there isn't that
bloody c2 stuff to get into the way.  (And it's still in kithrup's kernel,
although I, as a user and administrator, don't really see it because of the
sls that "fixes" it.)

SysVr4 is All Things to All People.  And a lot of that functionality comes
from the kernel, which is most things to most people.  This makes the kernel
larger and slower (usually; it's possible to optimize it as well as getting
rid of some of the cruft); it definitely adds maintainance problems, as
there is so much more to maintain.  Being a systems maintainance person,
such things frighten me.

>V.4 was pretty stable back then.  I haven't poked at a more recent version.
>ODT (which, note carefully, was young then) was somewhat less reliable than
>V.4.

I just rebooted kithrup, because I thought there was a network problem
(turns out someone next door was trying to splice in an ethernet cable, and
didn't tell me; c'est la vie).  Other than that, kithrup is a very stable
news, mail, and network machine.  (I don't do a whole lot of development on
kithrup right now, because I have such neat toys at work to play with.
Note, though, that I got cnews and nntp all working on kithrup, with only a
little bit of work.  I forget which compiler I used, though; I think I ended
up using /bin/cc [instead of gcc] because I trusted it more.  [I know msc
very, very well by now, I only know gcc fairly well.])

I've had kithrup up for 3 months at a shot, running in 4Mbytes with only 122
megabytes of disk space; since I rarely let free disk space drop below
30Mbytes, that means that I used up very little disk space for my OS.
("Only" about 40Mbytes, I think, although it might have been as much as 60.)
(Note that you should add about another 10Mbytes or so for the devsys, since
I tend to install the whole thing and then remove portions later. Silly
me.)

3.2.0 and 3.2.1 (aka ODT) were not very good, I honestly believe, and I have
never been terribly quiet about it (somewhat to the detriment of SCO, I must
admit 8-)).  3.2v2 is a *nice* OS.  It's stable, for me anyway, and has most
of the features I want.  (Well, I want long filenames right now, but that
should be available soon.)  *I* don't want or need all of what SysVr4 has to
offer.  People (myself included) have complained bitterly about sco's c2
cruft, because, although they admit there are people who want it and
situations where it would be a good thing, it's *not* an option, it's
*always* there.  I feel the same way about about 40% of what SysVr4 has to
offer:  some of it I would like, the rest I don't really need.

(For example:  why do the x.out and COFF compatibilty handlers have to
reside in the kernel?  It would be very nice if they could be elsewhere.
Kinda like the '286 emulator in 3.2, /etc/emul286 [I think].)

Anyway, those are just my opinions.  Time to go to bed.

-- 
Sean Eric Fagan  | "I made the universe, but please don't blame me for it;
sef at kithrup.COM  |  I had a bellyache at the time."
-----------------+           -- The Turtle (Stephen King, _It_)
Any opinions expressed are my own, and generally unpopular with others.



More information about the Comp.unix.sysv386 mailing list