[sf-lug] Has Linux outgrown Unix?

Rick Moen rick at linuxmafia.com
Wed Oct 15 17:47:16 PDT 2008

Minor correction (not that it's important):

> The whole BSD v. SystemV thing pretty much ended in 1988 with USL and
> Sun's release of System V r. 4.0, which merged the most compelling
> advantages of the Berkeley and New Jersey variants of that day.
> Everyone got aboard except for the *BSD people -- who grudgingly did
> later, anyway, courtesy of SUS and (especially) POSIX.

Of the BSDs, only Darwin / OS X has bothered to get Single Unix
Specification certification.  (For one thing, it costs real money[1],
and, since few people care about SUS, {Free|Net|Open|DragonFly}BSD
have not thrown that particular money down that particular rathole.

Being the product of an IEEE program (IEEE 1003), POSIX.1 is less of a
cash cow (http://posixcertified.ieee.org/posix-fee-schedule-1.1.pdf) --
$5000-$12000 as a one-time fee, instead of a big fee up-front plus
annual trademark licensing fees.  Still, none of the non-Apple BSDs were
ever certified -- but are mostly compliant, that being a good functional
definition of what it means to be an *ix, these days.

So, of course, is Linux (mostly POSIX.1 compliant), and the way it got
there is instructive -- in particular, as to how Linux helped put an end
to meaningless standards wars:

Basically, from the earliest days of Linux, corporate people kept coming
to Linux and saying "We need full POSIX compliance!"  And, his bullshit 
detector having been triggered, he would say, "To do _what_ with?  If
there's a POSIX call that your application is _actually using_ that
isn't yet serviced by the Linux kernel, post details and we'll code it."

In that way, over a couple of years, Linux gained (only) the parts of
POSIX.1 that are _actually used_, a relatively sparse approach that
prevented it becoming a steaming pile of... code like HP-UX, AIX, et

And everyone has seen the operational advantages of coding only what
people actually need, instead of taking marching orders from a standards
committe.  Which is why nobody with half a brain cares, these days,
about full POSIX (or SUS) compliance.

[1] http://www.opengroup.org/openbrand/Brandfees.htm

More information about the sf-lug mailing list