This is pretty mind-blowing stuff, but the paper doesn’t really hammer the point home until you read the man pages:
http://man.cat-v.org/plan_9/4/
Look for any of the entries ending in fs.
(I also wish we had an “operating systems” tag.)
This is pretty mind-blowing stuff, but the paper doesn’t really hammer the point home until you read the man pages:
http://man.cat-v.org/plan_9/4/
Look for any of the entries ending in fs.
(I also wish we had an “operating systems” tag.)
A quip I saw somewhere: “If we only spent half as much time maintaining Plan 9 as we did talking about it, it might have taken off.”
Find it interesting? Grab a 9front ISO, and try to use it for something. There’s a huge amount lacking, but a huge amount is also there and usable, and I do use it regularly (mostly for the sake of hacking on Plan 9/running some of my own code on it).
My theory–and I’d love to be corrected–is that Plan 9 failed because AT&T/Alcatel failed to open-source and promote it as a viable operating system.
It would’ve been such a better answer than Unix, Linux, or anything else. :(
I don’t think that’s it. Unix didn’t take off because it was open sourced (it was the opposite, and forcefully so); P9 arguably had just as much innovation in it as Unix, if not more in a number of ways. I think what really happened is all of our eggs were counted in the Unix basket (MINIX, BSD, Linux, etc etc etc), and moving those resources over for what would be a total change in architecture wouldn'tve seen feasible
That was Dennis Ritchie’s assessment when it was first released as a commercial product in 1995:
I don’t know enough about the 1995-1999 period to know what the odds of Plan9 doing well in the embedded space would’ve been, or why it didn’t succeed there. It does seem to have come out in time to plausibly have beaten Linux to the market. In 1995 embedded Linux wasn’t really usable, and afaict the first products using it came out in 1999.
The more interesting push to make it a product was Inferno, which added a bytecode layer, and positioned the whole system as a kind of ultra-portable, distributed computing architecture rather than a traditional operating system. It had a lot of novel ideas, and arguably did “write once, run anywhere” better than Java, with a stack for portable, bytecode-compiled, distributed software and services that’s more elegant than the whole Enterprise Java stack. But it never really succeeded in getting customers. Java was targeted and marketed better, for one thing, focusing on the web and enterprise servers, which was a huge market that picked it up. While Inferno was focused on a distributed-and-embedded market that didn’t exist / didn’t really find interest in it. Embedded people seemed not to care that much about distributed-systems at the time and by the turn of the century were just using Linux. Maybe if the Internet-of-Things had taken off in the ‘90s, Inferno would’ve gotten customers there.