The project is mostly dead since I haven't had the time to work on this in years. If anyone's interested in picking it up and porting it to the latest kernel version, I'd be happy to help in any way I can!
The interesting bits are mostly in the a.out (plan9's executable format) loader that enable implementing the small set of plan9 system calls.
One of the most interesting aspects of Plan 9 is the lack of a root user. Did you ever have any thoughts on the possibility of implementing that in Linux?
The goal was to bring Plan 9's most useful tools to Linux, rather than modify Linux to become more Plan 9-ish. Glendix doesn't modify the core or essence of Linux in any meaningful way (it's more like Wine in that sense), so the question of removing the root user never came up.
Couple of sub-projects within Glendix that are pretty interesting though are the synthetic /net filesystem that serves a TCP API, and per-process namespaces. These are simply in addition to what Linux already provides, so they don't impact existing Linux applications in any way. Glendix runs just like any other Linux installation, you just have access to more Plan 9 stuff.
The root concept is replaced by capabilities (sort of). So, for instance, the process owner that boots the machine (usually called bootes) can kill processes and spawn new ones as any username. That doesn't mean that they can fake your credentials to the network though, you need to authenticate with the auth server which will be on another machine. That you can access your files on the fileserver.
The concept of a network of connected islands is replaced by a network of connected services.
I'm just requesting a common bit of content on web page; it wasn't really anything to do with HN.
In this particular case my immediate thought was what is its relationship to plan9port and a date would have helped.
All web pages for projects, codes, blogs, should have a bunch of fairly standard things such as dates, news, about, source url, and a bunch of other stuff.
plan9 (and Go) people seem to have a bit of disdain for dates and version numbers which I actually find admirable in a way. Heroic and futile but admirable :)
I did this a few weeks ago on an OS X machine and it worked wonderfully. I installed it to give Sam a try, as I am learning Ed and wanted to compare them.
I had once hoped that Google would resurrect Plan 9, since they have so many former AT&T employees on staff. All it really needs it modern drivers, a wireless stack, and a browser (which they have with Chrome). Alas, it is not to be.
I've never used it but I can't imagine it has a browser. The two surviving browser cores (webkit/blink and gecko) are both utter nightmares of C++ and/or XUL, neither of which has any runtime support on plan 9.
Plan 9 from user space is a port of userspace programs to other hosts. This project is an attempt to make the host supportive of the userspace programs, rather than the other way around. Also, plan 9 from user space is not abandoned.
The project is mostly dead since I haven't had the time to work on this in years. If anyone's interested in picking it up and porting it to the latest kernel version, I'd be happy to help in any way I can!
The interesting bits are mostly in the a.out (plan9's executable format) loader that enable implementing the small set of plan9 system calls.