Re: [tlug] Peeling onions.

[Uva Coder <uva.coder@example.com>]

On Fri, 4 Mar 2005 14:28:03 -0500, Josh Glover <jmglov@example.com> wrote:
> 
> While I would agree with you whole-heartedly on LILO, I actually think
> GRUB is quite nice. Elegant design within the limitations imposed by
> each architecture, very extensible. What makes you see a train wreck,
> specifically?

GRUB didn't work for Plan 9 until last week. The Plan 9 folks added
GRUB support to the kernel. This is a little bit of the tail wagging
the dog, but some folks want to use GRUB; what can I say. ;-)

For those that might be interested in Plan 9 kernel history, there is
a new site that displays Plan 9 develop over its lifetime. See:
http://swtch.com/cgi-bin/plan9history.cgi?v=html;w=main
  
> Isn't this the Unix paradigm, though? I thought that Unix provided
> these different IPC methods so that the messaging protocols could be
> deferred to the application level. While I see the power of a system
> protocol like P9 (please excuse my blatant assumptions here about what
> P9 does--I am just inferring from context as I go here, being ignorant
> of Plan 9's design and implementation), I also understand why you need
> generic IPC as well. (Again, I know nothing of Plan 9, but knowing the
> way the BL chaps think--or at least, I *think* that I know how they
> think, having devoured all of the books that they have written--I
> imagine that it does both).

No, Unix edition 8 and later paradigms were not widely published nor accepted.
The current Linux paradigm is a combined Unix 7 edition and BSD paradigms; 
actually heavy on the BSD. 

Unix 8th, 9th and 10th edition had a lot of interesting programs and a
new ideas concerning what should be a working environment. Plan 9
borrows heavily the ideas behind the later editions. So to say that
Plan 9 does look or behave like Unix is not quite correct. But know
you can see why the claim "Plan 9 is how the original developers of
UNIX dreamed UNIX should be" comes about; or even perhaps Rob Pike's
claim that "Unix is dead and beginning to smell".

There are many things that didn't make into Plan 9 from later Unix
too. Plan 9 is a bit minimalistic. Plan 9 doesn't come with a Fair
Share Scheduler. Currently, Plan 9 is a distributed working
environment, not a Distributed Operating System. Once a Fair Share
Scheduler is in place, perhaps we could call it the later.

> I would love to find out more. Is there a document on the order of
> 10-25 pages that hits the highlights of Plan 9's design? If so, please
> point me to it.

There are a lot of papers. Perhaps these pages will help: 
http://plan9.bell-labs.com/sys/doc/
http://plan9.bell-labs.com/sys/doc/9.html

Again, it would be nice to see Linux adopt many of the ideas. It would
seem as a bit of a Unix homecoming. Unix with out X can still be a
great working environment; especially if you can export the GUI (i.e.
rio) to any system across the internet.
 
> Agreed. And, if I am not totally off-base, that is one of the biggest
> ideas behind Plan 9, right?

Well, combined with use of namespaces then yes. But even if you didn't
have namespaces, having a single utility to handle files is a great
idea. It would alleviate many problems and reduce application sizes.
Applications wouldn't have to create specialized utilities to handle
the files and files types themselves but would have a centralized way
to handle them; thus imho reducing problems.
  
> Au contraire, mon frere (rhyme intended, give me a cookie)! I am
> finding this discussion one of the most interesting that we have had
> in these parts in many moons. Hell, I think SJT and wileyc have posted
> more in this thread (and the related ones) than in the past year
> combined! :)

Thanks, I'm glad that it interests you. Sorry that I'm not more
prolific today, I suffering from a hangover. I'm a bit reasoning
impaired today. ;-)

-uva