Something to ruffle some penguin feathers: The Unix Hater's Handbook

I Cast Fist@programming.dev to Linux@lemmy.ml – 67 points –
web.mit.edu

This is a 1994 book about the many woes that Unix derived systems brought to sysadmins that were used to other solutions. Considering the number of commands that Linux still uses, it's definitely worth a read.

17

You are viewing a single comment

I've been reading this for a few weeks now.

There's a section talking a bit about the history of unix that I quite liked. I'm young and don't know the whole story. According to this handbook, unix was used and taught in universities, so when students went out into the world they preferred unix systems. As a result, tons of companies started offering their own versions of unix touting portability with the other unixes but with their own differentiating extensions which sort of contradicts the claim of portability. People would write for one system and frustratingly find that their software doesn't work on a different vendor. I'm curious where linux fits into the story. This must have been before linux because the handbook makes no mention of it. Did linux and its free software license immediately kill off these other unixes? How much role did copyleft play?

That was the Unix wars of the ‘80s.

Linux started in ‘92 as a hobby project to create a x86 desktop Unix clone. Most Unices were tied to expensive proprietary hardware which most people couldn’t afford, but x86 equipment was fairly common.

There was 386BSD project which had the objective of porting BSD Unix to x86, but they were sued by AT&T. In the end, AT&T was using more BSD code then BSD was using Unix code.

The lawsuit chilled use of BSD code, and a young CS students decide to write a kernel from scratch rather then fork a BSD.

About the same time Linux was first released, IBM was looking for a Unix-like OS to sell on its x86 servers. The idea was, companies would get started on the low cost x86 servers and graduate to the expensive Power AIX servers. Linux fit the requirements, and it was under the GPL which meant competitors would have to release any changes they distributed to clients as a bonus.

Linux did not immediately kill the propriety Unixes. It wasn’t until after the dot com bubble burst (~2000) that Linux really started taking market share. The tech companies needed to shed expenses, and an easy way was to ditch the expensive Sun equipment running Solaris in favor of commodity x86 running Linux.

The GPL played a role since it meant people distributing Linux needed to release their changes. Linux distros can be fairly different though, so I’m not sure how much of a part it played.

I had a brief opportunity to play around with an AT&T workstation running unix, more like a 386 than any sun/sgi machine that costed as 2 new sportscars. It had a very brief life, despite of the quality of the box, it was pretty useless. Slow as hell windows 95 would run circles around it.

Then out comes DEC/Alpha with Dec's unix, was it ultrix? And in those machines later windows NT was also ported, so it was a testbed between the two worlds. Then RHat CDs rained on us
@jollyrogue @eah

Wikipedia says Ultrix was VAX, and OSF/1 and Tru64 Unix were Alpha.

I am almost certain the first system alpha was ported to was ultrix, those other ones didn't exist yet. Probably developed for alpha, but on its pre-release demo I saw it was ultrix. Sometimes I confused ultrix with sgi/irix
@jollyrogue

My first Linux installation was done using Red Hat CDs that I purchased for around $20. Probably around 1996. Patching was difficult. Drivers for many pieces of hardware didn't exist. Remember Plug and Play was pretty new at that time frame. Lots of manual resolution of things like driver interrupt conflicts (boards had physical jumpers that you could move to change which IRQ they asserted). Looking back on it, I can't believe any of us were doing it. But the eventual payout was wonderful. I can't imagine what 1996 me would think about how easy something like the latest Ubuntu is. I would probably be pretty awed because I have a decent understanding of the massive amount of work that has been poured into the ecosystem now to make it what it is today.

All that said, I will always have a soft spot for Solaris on an Ultraspark. That shit worked great.

Linux was in its infancy back then, being "just another flavor" of Unix with "little to tell it apart" (other than being free, like 386BSD). Linus Torvalds began working on his flavor of Unix because MINIX (a micro variant of Unix, aimed at students) licensing didn't allow one to freely distribute it. He also changed internal programs to GNU alternatives wherever possible. The GNU Project was already rolling since 1983, so it already a significant quantity of software written.

Another thing to keep in mind, the internet was still very small, connection was slow and expensive by the time the book came out. The majority of software distribution happened with physical media, like floppy disks and, later on, CDs. Mailing lists and usenet forums could instantly send messages back and forth, but uploading and downloading stuff was severely limited.

So, Linux didn't immediately kill commercial Unix (MacOS has Unix roots, some big companies might still use Solaris), but it being completely free allowed it to "evolve and proliferate like a virus", much like Unix before it.

Can confirm, still use Solaris and AIX. They're declining though in favour of Linux.

I'm curious where linux fits into the story.

Someone here on Lemmy recently shared a really interesting article about the invention of Linux, written by a friend of Linus Torvalds who also worked on the project for a long time: https://lwn.net/Articles/928581/