[TUHS] Why BSD didn't catch on more, and Linux did

Theodore Ts'o tytso at mit.edu
Thu Feb 8 01:13:47 AEST 2018

On Tue, Feb 06, 2018 at 08:29:27PM -0500, Clem Cole wrote:
> Like you I started with the bits from Linus and it was a little difficult -
>  lot of DIY - column A, tab B, update this.  Then I discovered the first
> full ``distro'' that seemed to make sense (Slackware) - which in fact was
> similar to BSD at the time (used V7/BSD conventions) and it mostly worked.

The lots of DIY period was between 1990 and 1992.  At that point we
used HJ Lu's boot/root diskette images, and you then manually
installed various software packages (emacs, gcc, etc.) by untarring
compressed tar files.

The very first distribution was MCC Interim Linux (from the University
of Manchester) in February 1992.  In May 1992 the University of Texas
released the TAMU release, and Peter MacDonald released the
Softlanding Linux System (SLS).  In December 1992 Yggdrasil released
their first relase that had X Windows.  And Slackware forked off from
SLS in mid-1993.

> In fact, BSD/386 at that point had a better install and that would get even
> better after the suit ended with the fork that Jordan Hubbard and Co did,
> but I think that was good.  As the 386 installs for a BSD got better, it
> pressured the Linux guys to make their stuff even better.  And by that
> time, Linux pretty much had parity on the kernel side, if not started to
> get the lead [Linux supported modules early on, which I think was a
> technological development that is overlooked but was huge in making Linux
> flexible when it needed to be].

386BSD 0.0 was released in March 1992, and BSD/386 first release
(0.3.1) was in April 1992.  Both of there were *after* the very first
Linux distribution from the University of Manchester.  From what I can
see, the reason why the Linux distributions were getting batter was
competition from each other: MCC, SLS, Yggdrasil, Slackware, etc.

I certainly see anyone who was saying "ooh, we need to improve our
distro's installer because 386BSD or BSD/386 is competing with us".
Instead, various Linux enthusiasts were posting reviews comparing the
ease of installation of MCC vs SLS vs Slackware, etc.

> BTW: lets not forgot the larger issue.   At this point the
> 'better' kernel technology is in Solaris, Tru64 et al.. but that's not
> running on PC technology.  Few people can afford such a machine for
> themselves.

Indeed.  From http://www.freebsddiary.org/linux.php, "Why is Linux
Successful?  An Opinion", published at Uniforum NZ in April 1999:

    "Linux has always had a pragmatic view of hardware, whist the BSDs
    carried a purist view. When I got my first 386 I had MFM style
    disk drives. At that the BSDs only supported SCSI. Now SCSI is
    undoubtedly the correct choice, however it did not match the
    common hardware profile out in the market. Linux had the advantage
    for the first three years that I ran it of supporting a more
    diverse range of hardware than the BSDs. The BSDs assumed you had
    purchased a machine to run a Unix-style OS on, while Linux assumed
    you had a machine and wanted to try Unix. Linux was much
    friendlier to someone just wanting to dip their toes in the
    water. In this respect Linux did something that the BSDs were
    unable to do to any great degree - grow the Unix user base."

> But then it comes out, the suit was not about copyright, but trade
> secrets.  We all had been 'mentally contaminated' by the AT&T IP at our
> respective colleges and universities.  So the court does do the right
> thing, and AT&T loses the case.

Well, MIT actually had multiple AT&T Licenses which did *not* have the
mental contamination clause.  That was because MIT steadfastly refused
to allow MIT students and staff to be contaminated, and MIT had more
friends in high (low) places at AT&T than USL's licensing department.
So although the USL's Licensing division was quite frustrated about
this, MIT was able to get research grants that included full Unix
source licenses all of MIT without the mental contimnation clause.  So
that means all of the MIT Project Athena Alumni who first looked at Unix
sources at MIT would not have been so contaminated.

In addition, given that the POSIX.1 1990 specifications was available,
a lot of the more advanced system calls were implemented from that, it
seems.... dubious that AT&T would have been able to claim that an
ANSI/ISO Standard was somehow a trade secret.  Given how ridiculous
lawyers can be, I'm sure they could *make* such a claim.  But whether
it would have been laughed out of court is a different question.

     	      	      	       		    - Ted

More information about the TUHS mailing list