Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 17 Nov 1996 11:13:28 +0100 (MET)
From:      Greg Lehey <grog@lemis.de>
To:        dyson@freebsd.org
Cc:        chat@freebsd.org (FreeBSD Chat)
Subject:   Re: "free" SCO O/S
Message-ID:  <199611171013.LAA25485@freebie.lemis.de>
In-Reply-To: <199611161734.MAA00747@dyson.iquest.net> from "John S. Dyson" at "Nov 16, 96 12:34:09 pm"

next in thread | previous in thread | raw e-mail | index | archive | help
John S. Dyson writes:
>>
>> Ok, so I get a prompt for the BTLD disk, and insert it.  Only it
>> appears that the 'slha' driver isn't on the disk.  Or perhaps I
>> mistyped it.  At any rate, this section from the SCO manual is
>> pertinent :
>>
>>            NOTE  If there are any errors during this extraction (linking)
>>            process, the process is aborted and you are forced to reboot.
>>
>> Like hell I am.  Stick it back on the shelf and worry about it some
>> other time.  Anyone tells me FreeBSD is difficult to install is going
>> to get laughed out of the room.

Too true.

> You need to get the BTLD disk from the symbios site.  It took me
> a long time (a few weeks of picking away at it) until I found this
> out (since I don't know anyone running the turkey.)  

I was better off.  I had already installed SCO 2.0 a hundred times
from floppy.  At least that came with documentation.  As Mike reports,
the 12 page brochure is mainly advertising and has to be the worst
description of how to install an OS that I have ever seen.

> Once I got it running, I was disappointed.  The darned thing is
> "license-manager" city.

Right.  What really pissed me off is that they sent me the CD and
something like an invoice (pretty primitive for the world's largest
selling UNIX), and then expected me to go across the web (which costs
me $$$) to register.  What if SCO was the only machine I had that
would connect to the net?  This is just plain *stupid*.

> It also has lots of bogus symbolic links into wierd places for the
> various system binaries.

Forgive them.  They've only just got a version which supports
symlinks, so they're showing off.  Other System V vendors have this
same love of symlinks.  To quote somebody who might prefer to remain
anonymous:

> it's really incredible. I understand using them to hide a specific rev,
> or to make a mount system that isn't radically affected by disks moving,
> but it's really gotten out of hand.
>
> in MOST cases a single symlink at a high level can hide a revision,
> yet these weenies today seem to think they should keep the top level
> the same and symlink every freakin' thing at the bottom of the trees ;-{
>
> which is exactly what they did with the C30 compiler. /usr/lib/cmplrs/cc
> is a symlink to /opt/318env. in /opt/318env/usr/bin each executable is
> a symlink to a file in /usr/lib/cmplrs/cc3.18, which is itself a symlink
> to /opt/318env/bin. for example, in /opt/318env/usr/bin we have:
>
> lrwxrwxrwx 1 root other 28 Sep  5 10:31 DBX -> ../usr/lib/cmplrs/cc3.18/DBX
> lrwxrwxrwx 1 root other 34 Sep  5 10:31 abicc -> ../usr/lib/cmplrs/cc3.18/abi/abicc
> lrwxrwxrwx 1 root other 34 Sep  5 10:31 abild -> ../usr/lib/cmplrs/cc3.18/abi/abild
> lrwxrwxrwx 1 root other 27 Sep  5 10:31 ar -> ../usr/lib/cmplrs/cc3.18/ar
> lrwxrwxrwx 1 root other 31 Sep  5 10:31 as -> ../usr/lib/cmplrs/cc3.18/driver
> lrwxrwxrwx 1 root other 29 Sep  5 10:31 btou -> ../usr/lib/cmplrs/cc3.18/btou
> lrwxrwxrwx 1 root other 31 Sep  5 10:31 cc -> ../usr/lib/cmplrs/cc3.18/driver
> lrwxrwxrwx 1 root other 29 Sep  5 10:31 cord -> ../usr/lib/cmplrs/cc3.18/cord
> lrwxrwxrwx 1 root other 28 Sep  5 10:31 dbx -> ../usr/lib/cmplrs/cc3.18/dbx
> lrwxrwxrwx 1 root other 28 Sep  5 10:31 dis -> ../usr/lib/cmplrs/cc3.18/dis
> lrwxrwxrwx 1 root other 29 Sep  5 10:31 kdbx -> ../usr/lib/cmplrs/cc3.18/kdbx
> lrwxrwxrwx 1 root other 27 Sep  5 10:31 ld -> ../usr/lib/cmplrs/cc3.18/ld
>
> note the ../, which takes us elsewhere yet again, to a tree of duplicate
> structure. care to venture a guess where cc winds up? don't bother.

> Ugly...  BTW, at least I could get FreeSCO to boot (after some
> agony.)  I had Solaris 2.5.1 for a week or so, and could never get
> it to work (and the friend who owns it couldn't either on his
> machines.)
>
> FreeSCO is not a stellar performer, but it does have much faster
> metadata perf (they have worked on the filesystem alot since I last
> used SVR3.X.)   Otherwise, it is a bit sluggish.

How much memory do you have?  I installed it on my old workhorse
486/66 with 16 MB, and it *crawls*.  Run FreeBSD on the same machine,
and it won't blow your head off, but there is an incredible
difference.

Another thing: the development system is all screwed up.  I'm trying
to prepare a CD-ROM of ported software for SCO, and I had the devil's
own job porting.  Things that would port out of the box on other
operating systems failed for hundreds of reasons.  Their only header
file which defines dev_t defines it as short, whereas in reality it's
long.  There are a couple of other things like that.  With the
possible exception of Tandem, it's the most painful UNIX development
environment I've ever encountered.

All in all, I wonder if SCO has done themselves a favour with this
idea.  The main advantages I see in SCO is that there is a lot of
software available for it, and that it comes with good documentation
(the $$ version really does, or did the last time I bought one).  The
documentation falls flat on Free SCO, and I wonder whether the
availability of applications is enough incentive to go through all the
pain.

Greg




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199611171013.LAA25485>