25 comments

[ 3.0 ms ] story [ 59.2 ms ] thread
17 years later:

Plan9 is going nowhere. That commercial spin off Vita Nuovo (http://www.vitanuova.com/) is quiet. The Linux and Solaris monolithic kernels are still strong.

Minix3 doesn't really progress (http://groups.google.com/group/minix3). The L4 microkernel is commercially successful (http://www.ok-labs.com/).

Does commercial success or popularity imply technically sound and superior ?
In the case of UNIX, it's not merely commercial success and popularity. It is the choice of a large portion of the smartest people working in technology. If we were talking about success amongst the hoi palloi (like, say, Windows), then your point would stand...but we're not. UNIX has seen its fortunes flounder and flourish multiple times over the span of 30+ years, and yet, today, UNIX runs the most important technologies on earth (Google, Amazon, the majority of all other web applications, the best smart phones, most smart devices with anything more powerful than a simple microcontroller, and a whole lot more), because the people building them chose UNIX.

So, is UNIX technically sound and superior? In the general case, I would say, unequivocally, yes.

It seems like almost all of this can be attributed to path dependence rather than technological superiority.
Path dependence in case of Google, Amazon, smartphones, websites? What kind of 'path' did they have behind them that would make them unix-dependent?
Kernighan and Mashey mentioned in "The UNIX Programming Environment" (1981): "Success or failure often depends on nontechnical factors, whose importance often goes unrecognized by those who evaluate systems on purely technical terms." And they go on to mention some of these nontechnical factors, such as having started out on the DEC PDP-11.

In Ken Thompson's Turing Award speech, he claimed: "I can't help but feel that I am receiving this honor for timing and serendipity as much as technical merit. UNIX swept into popularity with an industry-wide change from central main frames to autonomous minis. I suspect that Daniel Bobrow (1) would be here instead of me if he could not afford a PDP-10 and and had to "settle" for a PDP-11." http://cm.bell-labs.com/who/ken/trust.html

Similar fate apparently happened due to the specialization of the MIT AI Lab's ITS: "Of course, ITS wasn't portable by any stretch of the imagination. It was an operating system in the old school of development, and it died rather suddenly when DEC announced that they were discontinuing the PDP-10 and its descendants in favor of the PDP-11 and VAX systems." http://www.crackmonkey.org/unix.html

(Personally, I only use a UNIX because it's a tool of the trade. It's free, or built into Macs. I don't enjoy it, but I use it, and found the _Unix Hater's Handbook_ refreshing.) http://www.cs.washington.edu/homes/weise/uhh-download.html

In particular, the "X-Windows Disaster" chapter of the Unix Hater's Handbook should be required reading for anyone considering GUI work on a UNIX-derived platform--and is also absolutely hilarious.

An HTML version of the chapter is here: http://www.art.net/~hopkins/Don/unix-haters/x-windows/disast...

I would be happy to have a multi-architecture, network transparent, modular graphic environment nicer than X.

Do you have one? I certainly don't.

It's easy to criticize the work of others. Doing it better is a whole different game.

Eh, making light of something's flaws is not the same as denying its value or strengths. I haven't been a desktop linux user for 5+ years because I hate the X Window System.

Did you read the chapter I linked to? It really is quite entertaining.

I hate hardware.

Hardware is the thing that prevents microkernels with carefully distributed functions.

But hardware is still land where dragons are, therefore you have to do everything with it in kernel-space (it's unsafe and isn't delegatable mostly).

If hardware would soon die and we'd be programming on the turing machine, we'd get microkernel OSes in no time!

> If hardware would soon die and we'd be programming on the turing machine, we'd get microkernel OSes in no time!

That'd be da bomb.

> "in no time"

> That'd be da bomb.

Actually, it would be a singularity.

It's a joke: (one of) Turings enigma-decrypting machines at Bletchley Park was nicknamed "Bombe". (and it was a mechanic brute-force attack, far from a Turing-machine)
(comment deleted)
"Name a product that succeeds by running UNIX as an application." Hmhm, Xen? I think I'd name quite a few.
Cygwin, Interix (a.k.a. SFU) on Windows NT
Would you trust your application to run in production on a Cygwin machine?
Why not? Thousands of applications run on Cygwin just fine. There's no reason to mistrust Cygwin more than you mistrust, say, MFC.

However in most cases you can delpoy to a more optimal platform (actual UNIX).

I wouldn't trust my apps to run on a Windows machine, period.

Running them under Cygwin, at least makes them easily portable to more refined environments.

That existed in 1992?
I'm just showing that some points that were true in 1992 aren't now.