There are people who hate Unix. There has been a mailing list, and the book The Unix-Haters Handbook/Book and Barf Bag, which Amazon.com reports is out of print. It is now available online.
This book expresses fiery opinions, opinions that have been obsoleted by later developments, and even some pretty valid criticisms of Unix. Andrew K's review observes that some of the flaws persist on "proprietary" versions of Unix, but are able to be resolved on "free" OSes like Linux and the free BSDs due to the ready availability of source code.
... Your book is a pudding stuffed with apposite observations. Like excrement, it contains enough indigested nuggets of nutrition to sustain life for some... | ||
| -- Dennis Ritchie | ||
In a recent browse, my wide conclusion is that the criticisms presented tend to represent:
Problems that came out of design compromises, necessary at the time, that have gradually gotten cast in stone (e.g. sundry ugliness within X, NFS ). I would, for instance, agree that if there was not a need to keep NFS compatible with MS-DOS file access requirements, it could have been a much nicer system. (In effect, that makes the problems with NFS Microsoft's fault for having a shoddy operating system.) I also agree that xauth ought to have been constructed with rather a lot more attention to "ease of use."
Complaints that Unix doesn't have the "beauty of design" inherent to LISP -based systems.
Obsolete claims.
The authors write "Will journaling become prevalent in the Unix world at large? Probably not. After all, it's nonstandard. "
That is no longer true; most of the Unix implementations that remain commercially viable offer either their own journalling filesystem JFS, or integrate Veritas' journalling implementation. And on Linux , there are several journalling filesystems readily available.
They bash Sendmail for being pretty insecure. Which has historically been pretty fair.
There haven't been many Sendmail exploits published since about 1996 or 1997, which I think means that it has gotten somewhat cleaned up so as to be less vulnerable.
It's not of vast continuing importance, as newer alternatives like qmail and Postfix have architectures that (by running as special unprivileged users) eliminate the classes of security exploits to which Sendmail was vulnerable.
It is worth noting that these people haven't come up with any alternative systems that seem to be preferable to Unix (aside, perhaps, from ITS ). They promote certain design points of certain systems, but none represent reasonable replacements for the overall functionality of Unix systems.
Sun NeWS is cited as a technically superior alternative to X; unfortunately, licensing is important, and if X lost out on "quality of design," it won on the basis of being widely usable. NeWS might have been technically better, but if people won't buy it because they are worried about Sun's control, or due to expensive cross-licensing, then it fails, regardless of how good it may be.
The system seemingly considered nearest to the "right thing" is ITS; unfortunately, the last ITS system at MIT was shut down in 1990. The Hackers Dictionary ITS - Jargon Entry indicates that:
ITS worshipers manage somehow to continue believing that an OS maintained by assembly-language hand-hacking that supported only monocase 6-character filenames in one directory per account remains superior to today's state of commercial art (their venom aginst Unix is particularly intense).
Now available as a PDF document; note who is hosting it...
See also here...
Don Hopkins has been known to visit Linux-related newsgroups and make disparaging remarks about X. He is noted for his preference to call it X Windows, not out of ignorance, but rather for the deliberate reason that it it bugs the purists.
People incorrectly mistake his use of terminology and his lack of "proper worshipfulness" for ignorance of the power or the name of X. If you want to dispute about X with him, make sure you know your stuff. He may not like X, but likely understands it better than almost any of us...
The criticisms described in his chapter on X are largely still valid, and this is the one chapter of the book that most fairly characterizes a set of unaddressed problems.
Happily, Motif is mostly fading away in favor of GTK and Qt , and there has been, of late, some actual new X innovation taking place. Ten years late...
Note that with the availability of PDP-10 emulators, there is now no reason for disgruntled ex-ITSers to continue to be disgruntled. They can run emulated ITS atop, well, um, ... Unix...