Hacker News Reader

Comment by spijdar | Hacker News Reader

If you count 70s and 80s "Unixes" then on its face it is a bit strange, but a lot of 70s and 80s "Unixes" don't exactly resemble what we think of as "Unix" anyway.

If instead you think of SysVR4 as the first "Unix", then Amiga Unix was indeed a very early Unix. I think this is a useful distinction, because de facto most of the software interfaces we associate with "Unix" are just System V (especially R4) in a trench coat. Note that POSIX and and SysVR4 released the same year (1988); they're technically unaffiliated efforts but represent a consolidation of a bunch of competing ideas into a ... tacit compromise.

Or, being more practical, SysVR4 is the absolute oldest "Unix" you're going to have a good chance of building modern (1990-2020s) software made "for unix" on. You can get a surprising amount of mileage out of a SysVR4 distribution -- but go any older, and you'll be in for a lot of "fun"!

View on HN

> but a lot of 70s and 80s "Unixes" don't exactly resemble what we think of as "Unix" anyway

And that's exactly why the term "early Unix" suggests "pre-SVR4". Once a platform has matured, it's not "early" anymore.

The whole thing is weirdly written. For example:

> Like many early Unix variants, Amiga Unix never became wildly popular

Except SVR4 was popular.

So they're either saying Amix was early Unix, then the GP is correct that it wasn't early Unix. Or they're saying that SVR4 was unpopular, which is also untrue.

I don't think the blurb is intending to suggest either of these points though. I'm sure people maintaining a fan site for Amix would understand their history. So I just think they've written the blurb very poorly. Poor enough that the default conclusion people are likely to draw is a technically incorrect one.

I don't see how that's incongruent. It says many early Unix variants never became popular, not all early Unix variants.

My point is that if they’re talking specifically about SVR4, then it was popular. And if they’re not talking specifically about SVR4, then it’s not “early Unix”.

As I said, I’m not trying to claim that they’re “wrong”. Just that the whole thing is phrased poorly because it’s really not clear what their context is. And that’s easily demonstrated by the fact that we’re arguing over said context here.

> a lot of 70s and 80s "Unixes" don't exactly resemble what we think of as "Unix" anyway.

As someone who was a UNIX developer (both kernel and userland) working for a UNIX support shop (Interactive Systems Corporation, later bought by Kodak and then Sun) from the mid 70's, starting with UNIX 6, through the late 80's and once gave a Usenix talk called "Everything you wanted to know about System V but were afraid to ask", where I held up the white System III manual and the black System V manual and joked that they had gone to the dark side, I find this comment utterly nonsensical. I can look through today's BSD man pages, or its code, and it's very familiar.

> If instead you think of SysVR4 as the first "Unix"

But of course it wasn't.

I think the point is that many nowadays only think of GNU/Linux as UNIX, which of course isn't how it is supposed to be.

No, that wasn't their point ... that's a different wrong notion.

A lot of 90's stuff ran great on SunOS 4.x!

Yes, but SunOS 4 was both extremely popular (enough that a lot of software had explicit support for running on it) and implemented a decent amount of System V and POSIX compatibility!

Probably most notably, it implemented SysV shared memory (sys/shm.h) plus messages/semaphores, STREAM support, SysV termio, SysV libcurses, and probably others I'm not aware of.

I'm not sure how much any of these helped run software, but it bears pointing out anyway.

Very true. SunOS 4.x is still my favorite 90's Unix. I had a Sun 3 box for a while, then got a low end Sparc Station at home! Eventually in the late 90's I gave in and installed Solaris. 2.4 and earlier was kinda rough, but it was pretty decent by 2.5.

Reply on news.ycombinator.com