Plan 9 from Bell Labs (1995) • Hacker News Board

Re: Plan 9 from Bell Labs (1995)

#11

Post by jfb » Fri, Aug 17, 2012, 5:57 PM UTC

deno wrote: post #8

jfb wrote: post #2

I wish it had stuck. In a related vein, I idly wonder if there's a project around intended to bring useful security to a practical Unix derivative. Nobody uses a big timesharing box any longer (for all reasonable definitions of nobody); and the protections mooted to keep "users" away from "wheel" are basically pointless and counterproductive in a world where attacks come from outside. Apple's code-signing is one reac…

You can do quite a bit in Linux with CGroups and SELinux. Projects making good use of those are Fedora, Systemd and Chromium browser.

I'll check it out. Next up, doing away with POSIX.

Re: Plan 9 from Bell Labs (1995)

#12

Post by jff » Fri, Aug 17, 2012, 5:59 PM UTC

Plan 9 pops up on the front page from time to time, and I see a lot of people expressing a lot of misconceptions about it. I know there are at least 2 or 3 other Plan 9 users on this site, and they help try to clear up some of the questions.

I've done quite a bit of poking around in Plan 9, but of course I don't know all of it. I've hacked in the kernel and worked on the 64-bit version as well, I've fooled with the compilers, and I've hacked on the window manager.

If anyone wants to ask questions about some aspect of Plan 9, I'll do my best to answer--and if I can't, I bet others can jump in and take over. It would be great to clear up some of the misconceptions, myths, and FUD!

Re: Plan 9 from Bell Labs (1995)

#13

Post by jff » Fri, Aug 17, 2012, 6:02 PM UTC

jfb wrote: post #11

deno wrote: post #8

Earlier quoted context omitted.

You can do quite a bit in Linux with CGroups and SELinux. Projects making good use of those are Fedora, Systemd and Chromium browser.

I'll check it out. Next up, doing away with POSIX.

POSIX defines a pretty good framework, which is why Plan 9 is POSIX-ish. They changed things that they thought needed improvement. I just wanted to point out that doing away with POSIX merely to do away with POSIX isn't necessarily wise; if you don't have a better idea to use instead, you'll just end up reimplementing much of POSIX under another name.

Re: Plan 9 from Bell Labs (1995)

#14

Post by rbanffy » Fri, Aug 17, 2012, 6:09 PM UTC

jfb wrote: post #2

I wish it had stuck. In a related vein, I idly wonder if there's a project around intended to bring useful security to a practical Unix derivative. Nobody uses a big timesharing box any longer (for all reasonable definitions of nobody); and the protections mooted to keep "users" away from "wheel" are basically pointless and counterproductive in a world where attacks come from outside. Apple's code-signing is one reac…

> the protections mooted to keep "users" away from "wheel" are basically pointless and counterproductive in a world where attacks come from outside.

Attacks may come from the outside, but they usually exploit code running on the target. IIRC, Apparmor is an implementation of what you seek - an easy way to limit what any given executable can do, in addition to what its process credentials can do.

Re: Plan 9 from Bell Labs (1995)

#15

Post by sp332 » Fri, Aug 17, 2012, 6:29 PM UTC

mseepgood wrote: post #3

The linked text is from the 90s. Not exactly news. But Plan 9 was a big improvement over Unix. Unfortunately, good things are often ignored.

A lot of the ideas in Plan 9 were ported to Linux. UTF-8, the /proc filesystem, and the idea of union filesystems. Also Plan 9 is arguably better because of what's not in it http://c2.com/cgi/wiki?WhatIsNotInPlanNine

Re: Plan 9 from Bell Labs (1995)

#16

Post by codgercoder » Fri, Aug 17, 2012, 6:48 PM UTC

And the relationship to Inferno is...?

Re: Plan 9 from Bell Labs (1995)

#17

Post by bbguitar » Fri, Aug 17, 2012, 6:48 PM UTC

Phew, Nostalgia indeed. I looked a Plan 9 back in late 90's when we were doing research for alternative OS's and ways of sharing stuff securely across a network without the cruft of DFS...

It had a lot of promise. What did become of it I wonder?

Re: Plan 9 from Bell Labs (1995)

#18

Post by catenate » Fri, Aug 17, 2012, 6:56 PM UTC

jff wrote: post #9

scarmig wrote: post #6

Let me know when they release a decent editor for it...

As gchpaco points out, both sam and acme are pretty usable. Acme makes a good enough programming editor that I use it for all my coding at work, while I keep a tarball of the original sam port on a server for those times when I have to use a system with a crappy version of vi and no admin privileges to install something better. Oh, and you can run Vim on Plan 9 now. But it's kind of like visiting a foreign country an…

I used emacs for 11 years (93-04) and then acme-like editors (wily, plan9port acme, Acme SAC, Inferno acme) for 8 years (04-). Acme (especially guide files) made a distinct and pervasive change to my working style that emacs never did: sped me up, prevented me from forgetting, gave me help just when I need it, and reduced errors. Inferno's VM gave me a minimal unix environment whose executable and my scripts I can take via usb to clients' machines.

Re: Plan 9 from Bell Labs (1995)

#19

Post by juan_juarez » Fri, Aug 17, 2012, 7:01 PM UTC

mseepgood wrote: post #3

The linked text is from the 90s. Not exactly news. But Plan 9 was a big improvement over Unix. Unfortunately, good things are often ignored.

> Unfortunately, good things are often ignored

...which is why we're still using Unix.

Re: Plan 9 from Bell Labs (1995)

#20

Post by catenate » Fri, Aug 17, 2012, 7:05 PM UTC

codgercoder wrote: post #16

And the relationship to Inferno is...?

Inferno is kind of like Plan 9 running on a VM, with its own peculiarites like a really nice shell language. Having a VM, versus integrating with the host os like plan9port or cygwin, lets you use new features of the os like directory binding and the /env directory of shell variables. There's an os command to run programs on the underlying host, be it windows, windows with cygwin, or linux.