The Use of Name Spaces in Plan 9, by Rob Pike.
(via ericvh)
February 21, 2005
12 Comments »
RSS feed for comments on this post. TrackBack URI
Muli Ben-Yehuda's journal
Syndication
Latest from Twitter
Tweets by mulibyCalendar
Archives
- April 2023
- August 2022
- March 2019
- February 2019
- September 2017
- January 2012
- August 2011
- June 2011
- May 2011
- April 2011
- March 2011
- February 2011
- October 2010
- September 2010
- November 2009
- April 2009
- January 2009
- December 2008
- November 2008
- October 2008
- August 2008
- June 2008
- May 2008
- April 2008
- March 2008
- February 2008
- January 2008
- December 2007
- November 2007
- October 2007
- September 2007
- August 2007
- July 2007
- June 2007
- May 2007
- April 2007
- March 2007
- February 2007
- January 2007
- December 2006
- November 2006
- October 2006
- September 2006
- July 2006
- June 2006
- May 2006
- April 2006
- March 2006
- February 2006
- January 2006
- December 2005
- November 2005
- October 2005
- September 2005
- August 2005
- July 2005
- June 2005
- May 2005
- April 2005
- March 2005
- February 2005
- January 2005
- December 2004
- November 2004
- October 2004
- September 2004
- August 2004
- July 2004
- June 2004
- May 2004
- April 2004
- March 2004
- February 2004
- January 2004
- December 2003
- November 2003
- October 2003
- September 2003
- August 2003
- July 2003
- June 2003
Muli Ben-Yehuda's journal 
Interesting
I think someone should make a free clone of plan 9 😉
Comment by moshez — February 21, 2005 @ 4:45 PM |
Re: Interesting
The license (http://cm.bell-labs.com/plan9dist/license.html) appears to be OSI approved, so maybe it’s not *entirely* evil 😉
Comment by mulix — February 21, 2005 @ 4:50 PM |
Re: Interesting
The OSI disappoints me.
For example, if I give a CD with plan9 to someone for money (thus “Commercial Distribution”) and he sues Lucent, then they can sue me for any court damages, which it is left up to me to recover from the claimant. This pretty onerous — I doubt it will let Debian package this up (as redistributors would shy away from it like a hot potato). This is just from the terms I read — there seems to be a legal minefield in that license 😦
Comment by moshez — February 21, 2005 @ 5:00 PM
Re: Interesting
Yikes.
Comment by mulix — February 21, 2005 @ 7:07 PM
I am fairly certain I read this paper back in 2001 when I still had a Plan9 fetish.
I’ve since outgrown the fetish, but still enjoy the ideas surounding Plan9.
Comment by the_p0pe — February 21, 2005 @ 10:59 PM |
same here. i’d like to see some of the ideas extended even further than plan9 does, though… it seems they’re still limiting themselves in terms of the “sharing” paradigm…
why use a single fileserver, or a single cpu server? why not have -every- CPU on the network pool, with the local CPU taking precedence on local processes of course? same goes for disk space. am I completely off the ball on this? it seems entirely doable even maybe just with plan9 as it exists now and a bit of scripting / network auto-configuration.
Comment by reverius42 — February 22, 2005 @ 3:38 AM |
Optimizations by binary interfaces?
I noticed that a lot of the stuff in Linux /proc filesystem is similar to Plan 9’s use of namespaces. Any relationship?
The paper is from 1993. What happened to Plan 9 during the last 12 years?
Some operations involving file access and writing/reading textual stuff (like writing commands to the process’s ctl file) may be time-consuming and inefficient. Was any thought given to creation of streamlined and efficient binary interface, which is however based upon and is equivalent to the less-efficient interface? How (if at all) can this be done without destruction of the conceptual purity of Plan 9 filesystem and namespaces?
Comment by tddpirate — February 22, 2005 @ 10:16 PM |
HATE HATE HATE.
Why did they have to shoehorn everything into a hierarchical form?
Comment by taral — February 24, 2005 @ 12:12 AM |
Not everything – process manipulation (e.g. fork) is still done in the traditional way.
I found it pretty consistent; anything specifically you object to?
Comment by mulix — February 24, 2005 @ 1:06 PM |
Yes. Very few systems are actually trees. Dags, maybe, but not trees. Putting in a tree representation just annoys me, maybe just because it’s overdone. It is possible to represent a dag using a filespace-type-thing, but people then tend to think of fixed roots and such, which goes contrary to the whole spirit of the thing.
Comment by taral — February 24, 2005 @ 8:28 PM
Also, plan9 came awfully close to a nice object-capability system. But the lack of a clear method-invocation (read/write does NOT cut it) kind of ruined that.
Comment by taral — February 24, 2005 @ 8:29 PM
Plan9 is heartwarming. So much is done right. I love the way that the windowing system provides the same interface it consumes, literally providing a bunch of “virtual terminals” (windows) inside of the terminal where it is running. So that you can run it recursively. Or, because nearly everything is routed through the same filesystem protocol, you can graft on almost anything from a machine across the network—filesystem, mouse, terminal, etc.
It’s a bit weird to use, though. The notion of “users” is a bit different than in unix,… IIRC it is very difficult to set up a Plan9 machine and give people “shell accounts.” I got so far as setting up Plan9 running on bare hardware at the XCF. Vmware makes it incredibly easy to play with, tho.
Comment by nibot — April 15, 2005 @ 6:18 AM |