Minix 3.3.0

I&#x27;ve upvoted this because I would like to see minix succeed and for nostalgic reasons but I can&#x27;t help the feeling that minix has missed its window of opportunity. In practically every arena where minix could have established itself there are now formidable entities entrenched with mindshare, budgets and an installed base that minix can&#x27;t even begin to touch. I&#x27;ve got similar feelings towards plan 9.<p>Even so, minix has a spot, and that spot is in education, an OS that can function like a digital version of a petri dish, something that you can quickly morph to test out a new idea without having to drag a huge behemoth of a kernel behind you.<p>I&#x27;m quite sad about this, I think that if minix had been open instead of &#x27;published as a book&#x27; that minix would have been what Linux is today only better engineered and that we&#x27;d all be better off for that.<p>Not being GPL&#x27;d meant that a whole generation of hackers followed Linus Torvalds rather than buying a bunch of books from Prentice Hall. And so now we have 70&#x27;s era tech instead of 90&#x27;s.

The choice to use a NetBSD userland isn&#x27;t hugely surprising in light of the project&#x27;s focus on making everything as clean and portable as possible. Packages also make a lot of sense. NetBSD did something hugely ambitious with Pkgsrc--it is like FreeBSD&#x27;s ports, but was designed to build software in a coherent fashion on lots of different *nix platforms. At one point in time, I was using pkgsrc to build a somewhat complicated custom stack of open source software on AIX, compiling with IBM&#x27;s XLC compiler. Can&#x27;t say everything worked perfectly, but there was so much scaffolding already in place with pkgsrc that it simplified my process immeasurably.

I feel this document really captures the feeling that OS researchers had at the turn of the century:<p><a href="http://herpolhode.com/rob/utah2000.pdf" rel="nofollow">http:&#x2F;&#x2F;herpolhode.com&#x2F;rob&#x2F;utah2000.pdf</a><p>On one hand you had Windows NT taking over commercial space, on the other you had the Linux juggernaut gathering speed. There was not much left for the niche players.<p>Glad to see Minix survives still.

From a technological point of view, I think Minix looks really interesting. Especially the feature of a driver being able to crash, and not crashing the entire system. Security is also an added bonus.<p>But, it seems to me that it&#x27;s not really usable for most people, since driver support is very poor.<p>I would really like to know, if anyone knows the answer, how much effort it would take to create some sort of tool that converts a Linux driver to a Minix-compatible driver, or -- if that&#x27;s not possible -- how many man hours it would take to rewrite ALL Linux drivers to be Minix-compatible. Anyone know?

I&#x27;m suspicious that some people complain about dated technology and want something fresh not because the dated tech is obsolete or crippled but because they want to be an expert on new tech, starting from a level playing field and not have to catch up with people with a 20+ year head start.

Minix, the only *nix I&#x27;ve ever modified or recompiled.

There is only one question I ask to Minix: How much space and time does it take these days to compile a Minix kernel?<p>The current system I use takes about 220MB of RAM and about 15min on an underpowered netbook. (It is not Linux but still has decent hardware support.)<p>Of course I&#x27;m also curious what other folks who compile their own Linux kernels see as their &quot;minimum&quot; requirements.

AFAICS, Minix&#x27;s model will win in the end because security will be the most important OS feature in the future. It&#x27;s a matter of time before our life savings (crypto-currencies), all personal info, and even our lives (self driving cars, medical equipment) inevitably becomes dependent on the security of our computer(s).

I&#x27;m not understanding much from this thread&#x27;s conversation. I&#x27;m not so interested in OS development (anymore).<p>Yet I&#x27;m reading the conversation in full due to the great, knowledgable, and foremost <i>polite</i> discussions. Not having seen this in Hacker News in a long time...<p>Thank you all

Nice to see they have finally gotten to ARM support. I&#x27;ve worked on an MINIX3 ARM port for my thesis. Got really close to a running system but never got to finishing it after I got my degree.<p>Gonna try it out tonight!

I wonder if anybody uses minix in production anymore.

Has anyone benchmarked this? lmbench (I&#x27;m biased) or something else?

One way to spur adoption, let&#x27;s see a Yocto BSB for this

It would b interesting if minix could reuse the work done on mirageos by importing ocaml and build user space drivers in ocaml to be used by other programs.

Looks cool, but what is this &quot;printing&quot; and &quot;CD-ROM&quot; it is talking about?

This is real hacker news!