Owen Le Blanc: creator of the first Linux distribution
Ask a Linux enthusiast who created the Linux kernel, and odds are they will have no trouble naming Linus Torvalds—but many would be stumped if asked what the first Linux distribution was, and who created it. Some might guess Slackware, or its predecessor, Softlanding Linux System (SLS); both were arguably more influential but arrived just a bit later. The first honest-to-goodness distribution with a proper installer was MCC Interim Linux, created by Owen Le Blanc, released publicly in early 1992. I recently reached out to Le Blanc to learn more about his work on the distribution, what he has been doing since, and his thoughts on Linux in 2025.
![Owen Le Blanc [Owen Le Blanc]](https://static.lwn.net/images/2025/owen-le-blanc-sm.png)
When Torvalds first released the Linux kernel, he also provided "boot" and "root" disk images (intended for 5¼-inch floppy disks) to help users create a Linux system. There was not a proper installer, and users had to collect several other utilities to pull together a working system. This was not, even for the time, particularly user friendly. Clearly, Linux was going to need a little extra help on its path to world domination.
MCC Interim Linux
Le Blanc worked for the Manchester Computing Centre (MCC) at the University of Manchester from 1985 through 2016. He had been working with CDC 7600 and Cyber 170/730 machines, and eventually began working with HP 9000 workstations. Those ran Hewlett Packard's Unix, HP-UX, which made him want to learn more about Unix. He started with Xenix, but wanted something easier to use. He said that he also tried MINIX, but it was difficult to get working on a Intel 80486 CPU. When Torvalds announced Linux, Le Blanc decided to give it a try, and that too was a fair amount of work.
It was fairly hard to install Linux at first, because I think you needed both MS-DOS and Minix -- we used a bootloader from Minix before Werner Almesberger wrote LILO. You needed Minix because the DOS fdisk program could not write partitions for other operating systems, and you needed DOS if you wanted to do networking.
He was interested in improving the installation and created a Linux
system that could be installed from a pair of floppy disks. He made
use of ramdisk code by Ted Ts'o, took binaries from a proto-distribution
created by H. J. Lu, and he wrote the original Linux fdisk
"bugs and all
" for the project. According to the MCC
README for version 0.99.p10+, Le Blanc used his
distribution to install Linux on 386-SX machines
for the university's C and Unix courses. It was possible to install
Linux on 12 machines in about an hour, according to the
documentation, certainly an improvement over earlier installation
methods. Eventually, he asked his department if he could share the
work that he had been doing.
The department was reluctantly willing, provided (1) I gave them some credit for supporting its development, and (2) I made certain they had no responsibility for ongoing support for users. We agreed that calling it "MCC Interim Linux" suggested that MCC had developed it, and it was not intended to be a long term project.
The first MCC Interim Linux release (0.12) from February 1992, is, if not completely lost, certainly hard to find online. The 1.0 release is preserved on the ibiblio archive with other historic distributions such as Yggdrasil, Red Hat's "Mother's Day" 1.0 release, and SLS 1.03 and 1.05.
The distribution included things like GCC, GNU Awk, GNU Emacs, and the info packages because they were used in the university's courses. He did not attempt to include a wide assortment of other software, such as XFree86 (then X386), though.
Later releases can be found on debian.mcc.ac.uk,
in the non-debian directory. It has several releases, up to
the final 2.0
release from 1996. I had asked Le Blanc whether MCC had
been distributed commercially, and he said no one had ever sold copies
of MCC that he was aware of though "there were certainly people who
passed disks to others
". While doing some digging, though, I found
that Walnut
Creek had included MCC in some of its Linux Toolkit CD sets. For
example, the Internet Archive has the March 1997
set with MCC Interim Linux 1.0+, along with Slackware, Debian,
MkLinux, and others. Readers who have older Linux Toolkit CD sets, or similar
sets, gathering dust in their closets may have even earlier versions.
In all, Le Blanc said that the
project ran about seven years, with the last updates to the
distribution being released in early 1998. During the time MCC
was maintained, he said that he had received advice and suggestions
from many people, including Torvalds, Ts'o, Almesberger, Alan Cox, as
well as local users "especially John Heaton, Ted Harding and Nobby
Clark
".
The most valuable help was testing, which we tried to do before every release and update. In the early days I hoped I could find other people to develop add-ons to the system, but this never took off.
There is also some misinformation about MCC's development, he said, including
claims that it was developed by Bruce Perens. "Bruce did terrific work on Debian
and in many other ways, but he never had a connection with MCC Linux to my
knowledge
".
Le Blanc said that he used Debian from its early days, and eventually decided to encourage others to migrate to it as well. MCC Interim Linux included a migration tool for moving to Debian in its final release.
Supporting Linux
With other more ambitious and polished Linux distributions
available, Le Blanc focused his attention on managing university
servers and supporting Linux applications for MCC—later Information Technology Services
(ITS)—users. All of the servers ran
Debian, and he encouraged users to run Ubuntu—which he prefers
for desktops as well. He also supported Scientific Linux for the
computer science and mathematics department, but said that became harder and harder
to sustain because it was "always a bit behind for those who wanted
the latest versions of things
". He also continued to promote Linux
and said that he is still involved in open-source projects, such as LilyPond, today "but more as a
user and support person than as a developer
".
Despite being the home of the first Linux distribution, Le Blanc said,
the University of Manchester was always a bit reluctant to use open-source software
and he found himself "spending more and more energy to defend what we had
". He
said that there is still too much reluctance in universities, governments, and
businesses to use free and open-source software or anything that doesn't come with a
price tag. (LWN covered some of
hurdles for open-source adoption in Mexican government in March.)
This despite the fact that much of the commercial support for "industry standard" software is pretty awful. Of course, when you have systems making use of open source/free software components, then you need local expertise to manage and support them, and this goes against the belief that it's safest to have systems for which you can easily buy support, because they use out-of-the box components put together according to the manual. Although this approach produces poorer quality systems that cost more in the long run to maintain and update, it's very hard to convince (non-technical) higher managers of this.
In 2016, Le Blanc accepted voluntary severance from ITS and moved on to teaching, where he is still active.
Linux would not have made it this far without Le Blanc—and others like him—making incremental contributions that paved the way for more people to use and improve Linux. MCC Interim Linux did not last long, but it was an important stepping stone on the path to longer-lived distributions like Debian.
[Hat-tip to LWN reader Roger Whittaker for the article suggestion.]
Posted Apr 21, 2025 16:14 UTC (Mon)
by jepler (subscriber, #105975)
[Link]
I don't recall the specifics, but installing the system, gcc, and X from just a few floppy disks was amazing, even though once you'd done that there was nothing like a modern packaging system available to use.
Posted Apr 21, 2025 16:16 UTC (Mon)
by smoogen (subscriber, #97)
[Link]
We had been using a LynxOS Realtime Unix before this and having all kinds of problems with lack of working applications. A technician I worked with at the University said "You have got to try this new operating system!" and everything worked. I used it heavily until the connections from Socorro, New Mexico, US to Manchester, UK got too painful with failed FTP and other oddities. However I was hooked and have been using Linux ever since.
So thank you again.
Posted Apr 21, 2025 19:13 UTC (Mon)
by jkingweb (subscriber, #113039)
[Link] (7 responses)
Why was this so? DOS isn't exactly known for its networking capabilities, so I struggle to see the connection.
Posted Apr 21, 2025 19:35 UTC (Mon)
by corbet (editor, #1)
[Link] (2 responses)
Posted Apr 23, 2025 11:40 UTC (Wed)
by jkingweb (subscriber, #113039)
[Link] (1 responses)
Posted Apr 23, 2025 13:08 UTC (Wed)
by corbet (editor, #1)
[Link]
Posted Apr 21, 2025 20:47 UTC (Mon)
by hrw (subscriber, #44826)
[Link]
Simple API allowed to run bare networking stuff and/or TCP/IP stack.
I used those in 2000-2001 to write my master thesis (target machines were not good enough to run Linux).
Posted Apr 23, 2025 17:16 UTC (Wed)
by butlerm (subscriber, #13312)
[Link] (2 responses)
VAX VMS machines with DECNet and later TCP/IP were relatively common in some environments as well, although personal computers mostly accessed minicomputers and larger machines using terminal emulators at first, and usually over an RS232 serial port or (relatively slow) modem of some type. Even in Unix machines, UUCP over phone line was much more common than Ethernet for most of the 1980s. And on the Amiga, the Atari ST, and Apple II series, Ethernet interfaces were somewhere between rare and non-existent back then.
On the PC TCP/IP did not become relatively common until Trumpet Winsock was released for Windows 3.1 over DOS in the early 1990s or so, before Windows 95's and Windows NT's built in TCP/IP networking took over the world and supplanted most Layer 2 and Layer 3 ISO networking protocols in the 1995 timeframe. Before that the networking capabilities of most home computers largely consisted of dialing up Bulletin Board Systems using whatever POTS compatible modem was reasonably affordable at the time.
300 baud was common and relatively inexpensive by the early 1980s for that, with 1200 and 2400 bps soon to follow. 9600 baud modems were expensive at first, like cost more than your computer expensive, and often used for Unix and other large servers to do things like exchange email and USENET newsgroup messages over dialup several times a day back when long distance calls were cost prohibitive even for large institutions. Some BBS software usable by ordinary PCs had similar capabilities, and I believe Minitel made waves with X.25 terminals in France around the same time.
Posted Apr 23, 2025 18:03 UTC (Wed)
by Wol (subscriber, #4433)
[Link]
And actually, reading the article, a few things suddenly make sense ... apparently a ring is limited to 255 nodes, and it's coming back to me there was an American ring. I think I was at Pr1me Southampton, and they demo'd it by jumping from ring to ring. I know you usually typed "login username", but you could also type "login username on systemname", and istr some way of putting a gateway in - maybe like "login username on systemname via othersystemname".
Cheers,
Posted Apr 24, 2025 15:37 UTC (Thu)
by wittenberg (subscriber, #4473)
[Link]
Posted Apr 21, 2025 19:20 UTC (Mon)
by Cardinal_Bill (subscriber, #23688)
[Link]
Posted Apr 21, 2025 23:05 UTC (Mon)
by pabs (subscriber, #43278)
[Link]
Posted Apr 22, 2025 6:39 UTC (Tue)
by danieldk (subscriber, #27876)
[Link] (3 responses)
Posted Apr 24, 2025 16:05 UTC (Thu)
by Wol (subscriber, #4433)
[Link] (2 responses)
Cheers,
Posted Apr 24, 2025 17:14 UTC (Thu)
by amacater (subscriber, #790)
[Link] (1 responses)
Posted Apr 24, 2025 18:32 UTC (Thu)
by Wol (subscriber, #4433)
[Link]
I assumed it was a distro with a manual.
Cheers,
Posted Apr 23, 2025 22:39 UTC (Wed)
by rhowe (subscriber, #102862)
[Link]
He was demonstrating an early KDE release, probably more recent than I'd manage to build, on some pentium 2 (maybe 3?) desktops. I think it was probably 1999.
Needless to say, as someone who'd been using Linux for a couple of years at home on a 486 it had me sold and I applied to Manchester soon after.
You probably don't remember it, but I asked you if ftp.mcc.ac.uk was down in 2003 and you restarted the service, bringing it back. Happy, simpler days.
Posted Apr 24, 2025 1:53 UTC (Thu)
by irogers (subscriber, #121692)
[Link] (1 responses)
Posted Apr 24, 2025 20:23 UTC (Thu)
by Grimthorpe (subscriber, #106147)
[Link]
But one of my overriding memories of the machine room there was the NeXT cube that was used as the console device for the KSR supercomputer that allegedly could only run for up to 10 minutes before crashing.
Posted Apr 24, 2025 8:17 UTC (Thu)
by stsimb (subscriber, #805)
[Link] (1 responses)
It attracted people interested in Linux who lived in Manchester (I was a CS student there between '94 and '00) or even other places nearby (we regularly had Richard coming from Sheffield). We were about 20-30 people who attended semi-regularly. The meeting took place in a room inside the Manchester Computing Centre's building, and in order to reach it you had to walk through a corridor with an impressive view to the MCC's supercomputers (a Cray, a Fujitsu vector system and many more), all of which were part of JANET's resources shared amongst British universities.
In the meetings we had various speakers with interesting subjects to cover, and of course Owen & friends were always there to help. Their efforts created a thriving community and a forum for idea exchange among passionate individuals.
Thank you Owen, John, Ted and everybody else who was there at the time!
Posted Sep 27, 2025 16:17 UTC (Sat)
by cfillekes (guest, #179566)
[Link]
Posted May 2, 2025 11:52 UTC (Fri)
by boutell (guest, #177251)
[Link]
Posted May 2, 2025 11:52 UTC (Fri)
by boutell (guest, #177251)
[Link]
Posted May 4, 2025 17:58 UTC (Sun)
by cyberia (guest, #177290)
[Link] (1 responses)
Posted May 4, 2025 18:24 UTC (Sun)
by jake (editor, #205)
[Link]
ouch, indeed it does ... but now it goes here:
https://www.itservices.manchester.ac.uk/
which seems to the right place ...
thanks,
jake
I have fond memories of MCC
thank you
Stephen
DOS for networking?
Early versions of the kernel lacked advanced features like TCP/IP networking, so you had to use a different system to talk to the world.
DOS for networking?
DOS for networking?
I don't remember exactly when FAT filesystem support came in. I do remember, though, that GNU mtools was the lowest-friction way of dealing with diskettes and such for a long time.
Filesystem support
DOS for networking?
DOS for networking?
DOS for networking?
Wol
PARC was there first
My first exposure.
Thanks!
Great article!
Great article!
Wol
OT Linux Universe [WAS Great article!]
1994 was about the time when you could get Slackware 2.0 on one CD or a three or four CD box from Walnut Creek including Linux distributions and the tsx-11 software archive.
OT Linux Universe [WAS Great article!]
Wol
A true blast from the past, and a welcome one
Financial rather than "industry standard" motivation against open source?
Underused technology in Manchester
The NeXT cube's job for 99% of the time was to just show a colourful KSR logo spinning on the screen to anyone walking past.
Owen also organised and run ManLUG meetings back then
Owen also organised and run ManLUG meetings back then
Gratitude for MCC Interim Linux
Actually
"ITS"
"ITS"
> website of a department of a university in the USA with the same name