Computer Shopper – September 2019

ISSUE 379|COMPUTERSHOPPER|SEPTEMBER (^2019113113)
FROMNOT?SO?SMALLBEGINNINGS
Unix didn’t appear out of nowhere.Like many
great ideas, it evolved from somethingelse –in
this case,Multics, the MultiplexedInformation
and ComputingService.Developed by General
Electricand Bell Labs, under leadershipfrom
MIT,Multics was atime-sharingoperating
system designedforlarge public utilities,such
as electricitygrids and telephonenetworks.
It was one of the earliestmulti-processor
systems, designedforuse on specifichardware
that could be hot-swappedat will.
In many ways, it was years ahead of its
time.Multics was the first operatingsystem
both to provide ahierarchicalfile system and
to jettisonany restrictionon the length and
compositionof filenames.It’s easy to forget
that it took PCs another30 years –until
Windows95 and WindowsNT 3.5 –tosupport
equivalent flexibility;until then, filenames
were limited to an eight-character base name
and athree-character extension.
Take-up was enthusiastic,but Bell Labs
wasn’t happy. It thoughtMulticswas too
demanding,andtherequiredhardwaretoogreat
an investment.Five years after work began, in
1969,itwithdrewits developersfrom the team
and tasked them with buildingsomething
lighter in-house.Whatever theycame up with
would have to be less demanding,and able to
run on cheaperhardware.
Quitehowmuchcheapersoonbecameclear.
As Dennis Ritchie,who headedup
developmentalongsideKenThompson,
explainedto the LanguageDesign and
ProgrammingMethodologyconference in
Sydney, Australiain 1979: “Throughout1969,
we...lobbied intensivelyforthe purchaseof a
medium-scalemachineforwhich we promised
to writeanoperatingsystem; the machineswe
suggested were the DEC PDP-10 and the SDS
(later Xerox) Sigma 7. The effort was
frustrating,becauseour proposalswere never
clearly and finally turned down, but yet were
certainlynever accepted. Several times it
seemedwe were very near success.The final
blow to this effort came when we presented an
exquisitely complicated proposal,designedto
minimisefinancialoutlay, that involvedsome
outrightpurchase,some third-partylease,and
aplan to turn in aDEC KA-10 processoron the
soon-to-be-announcedand more capable
KI-10. The proposalwas rejected, and rumour
soon had it that WO Baker (then vice-
presidentof Research)had reacted to it with
the comment‘Bell Laboratories just doesn’t
do businessthis way’.”
BACKTO BASICS
So,lacking acomputer,work began on
blackboardsand paper instead. Thompson,
workingwith Ritchie and Joe Ossanna,
designedthe entire file system of what would
becomeUnics (the OS’s originalname) without
touching asingle computer.Indeed, Unics
remainedentirelytheoreticaluntil Thompson
discovered around-aboutroutetoasolution.
While workingon Multics,he’d developed
agame called Space Travel. It was simple by
modernstandards,but runningit on a
time-sharedsystem like Multicswas expensive.
So he looked foranalternativeplatform to
which he could port it. After settlingon aDEC
PDP-7,heand the rest of the team realised
that the same hardwarewould be asuitable
test-bed fortheir new operatingsystem.
Theycoded their rudimentaryfile system
using GECOS,runningon Multics,and
transported it to aborrowedPDP-7 using
paper tape.Then theyadded the commandline
ABOVE:KenThompson(sitting)and Dennis
Ritchie’s ground-breakingwork at Bell Labs resulted
in the developmentof Unix 50 years ago
UNIXAT50