[9fans] I found this discussion pretty funny

[9fans] I found this discussion pretty funny

[ron minnich]

get to the part about why it's so great that pipe is a system call,
not a device.

This is the sort of back-and-forth that reminds me why I can't quite
give up on plan 9 ...

http://yarchive.net/comp/linux/everything_is_file.html

"Do we create pipes by opening /dev/pipe? No. "

and

"Yes, some old-timers could argue that original UNIX didn't have sockets,
and that the BSD interface is ugly and an abomination and that it _should_
have been a namespace thing, but that argument falls flat on its face when
you realize that the "pipe()" system call _was_ in original UNIX, and has
all the same issues."

ah well. It seems that the guy who wrote pipe() might not agree that
he does not know much about Unix ... but so it goes.

ron

Re: [9fans] I found this discussion pretty funny

[erik quanstrom]

> "Yes, some old-timers could argue that original UNIX didn't have sockets,
> and that the BSD interface is ugly and an abomination and that it _should_
> have been a namespace thing, but that argument falls flat on its face when
> you realize that the "pipe()" system call _was_ in original UNIX, and has
> all the same issues."

appeal to the authority of the unix authors as a defense of sockets?
surely someone has missed a bit.

or another way, since the original was broken, we must not work correctly?

i'm tempted to draw one of those ascii slashdot memes

	unix →

	\o/
	 |
	/\

woosh!

- erik

Re: [9fans] I found this discussion pretty funny

[dexen deVries]

On Wednesday 17 of October 2012 06:40:45 ron minnich wrote:
> get to the part about why it's so great that pipe is a system call,
> not a device.
> 
> This is the sort of back-and-forth that reminds me why I can't quite
> give up on plan 9 ...
> 
> http://yarchive.net/comp/linux/everything_is_file.html
> 
> "Do we create pipes by opening /dev/pipe? No. "
> 
> and
> 
> "Yes, some old-timers could argue that original UNIX didn't have sockets,
> and that the BSD interface is ugly and an abomination and that it _should_
> have been a namespace thing, but that argument falls flat on its face when
> you realize that the "pipe()" system call _was_ in original UNIX, and has
> all the same issues."
> 
> ah well. It seems that the guy who wrote pipe() might not agree that
> he does not know much about Unix ... but so it goes.

a technical and organizational problem: back then Linux' /dev consisted of 
special files held on drive, its MAJOR/MINOR schema* getting dangerously 
cramped.

those days Linux' /dev is usually mounted at boot and maintained by kernel, 
somewhat alike /proc; adding some more entries probably isn't that much of a 
problem.


-- 
dexen deVries

[[[↓][→]]]

* MAJOR/MINOR being an unmountable namespace is quite ironic.

Re: [9fans] I found this discussion pretty funny

[Bakul Shah]

On Oct 17, 2012, at 6:48 AM, dexen deVries wrote:
> a technical and organizational problem: back then Linux' /dev consisted of 
> special files held on drive, its MAJOR/MINOR schema* getting dangerously 
> cramped.
> 
> those days Linux' /dev is usually mounted at boot and maintained by kernel, 
> somewhat alike /proc; adding some more entries probably isn't that much of a 
> problem.

You should explore /sys on Linux. They've embraced namespaces in a major way.

$ df /sys
Filesystem           1K-blocks      Used Available Use% Mounted on
sysfs                       0K        0K        0K   -  /sys
$ ls -lR /sys |wc
  10485   63451  419873
$ ls -l /sys/dev/block/8:1
lrwxrwxrwx. 1 root root 0 Oct 17 11:31 /sys/dev/block/8:1 -> ../../devices/pci0000:00/0000:00:0d.0/host2/target2:0:0/2:0:0:0/block/sda/sda1

Let us see where this leads:

$ ls -Ll /sys/dev/block/8:1
total 0K
-r--r--r--. 1 root root 4096 Oct 17 11:31 alignment_offset
-r--r--r--. 1 root root 4096 Oct 17 11:31 dev
-r--r--r--. 1 root root 4096 Oct 17 11:31 discard_alignment
drwxr-xr-x. 2 root root    0 Oct 17 11:26 holders 
-r--r--r--. 1 root root 4096 Oct 17 11:31 inflight
-r--r--r--. 1 root root 4096 Oct 17 11:31 partition
drwxr-xr-x. 2 root root    0 Oct 17 11:31 power
-r--r--r--. 1 root root 4096 Oct 17 11:25 size
-r--r--r--. 1 root root 4096 Oct 17 11:31 start
-r--r--r--. 1 root root 4096 Oct 17 11:31 stat
drwxr-xr-x. 2 root root    0 Oct 17 11:31 subsystem
drwxr-xr-x. 2 root root    0 Oct 17 11:31 trace
-rw-r--r--. 1 root root 4096 Oct 17 11:31 uevent

Exploring further...

$ ls -l /sys/dev/block/8:1/subsystem
lrwxrwxrwx. 1 root root 0 Oct 17 11:31 /sys/dev/block/8:1/subsystem -> ../../../../../../../../../class/block

Editorial comments left as an exercise to the reader!

Re: [9fans] I found this discussion pretty funny

[erik quanstrom]

> You should explore /sys on Linux. They've embraced namespaces in a major way.
>

what am i missing.  linux' /sys is just a synthetic filesystem.  where do you
see this exploring the idea of namespace?

- erik

Re: [9fans] I found this discussion pretty funny

[Bakul Shah]

On Wed, 17 Oct 2012 20:35:12 EDT erik quanstrom <quanstro@quanstro.net>  wrote:
> > You should explore /sys on Linux. They've embraced namespaces in a major way.
>
> what am i missing.  linux' /sys is just a synthetic filesystem.  where do you
> see this exploring the idea of namespace?

Oops.  s/namespace/synthetic filesystem/

Re: [9fans] I found this discussion pretty funny

[David Leimbach]

[-- Attachment #1: Type: text/plain, Size: 696 bytes --]

On Wed, Oct 17, 2012 at 6:10 PM, Bakul Shah <bakul@bitblocks.com> wrote:

> On Wed, 17 Oct 2012 20:35:12 EDT erik quanstrom <quanstro@quanstro.net>
>  wrote:
> > > You should explore /sys on Linux. They've embraced namespaces in a
> major way.
> >
> > what am i missing.  linux' /sys is just a synthetic filesystem.  where
> do you
> > see this exploring the idea of namespace?
>
> Oops.  s/namespace/synthetic filesystem/
>
> It'd be a lot better if it was a synthetic file system in a restricted
namespace.  I don't think 100% of the processes need to or should have the
opportunity to mess with /sys.   But maybe that's just all this stuff I
learned from Plan 9.  :-)

Dave

[-- Attachment #2: Type: text/html, Size: 1098 bytes --]

Re: [9fans] I found this discussion pretty funny

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 168 bytes --]

http://planb.lsub.org/ls/octopus.html

is a user interface that includes graphics, exploits name space
representations, and doesn't just mimic the   xerox desktop.

[-- Attachment #2: Type: text/html, Size: 246 bytes --]

Re: [9fans] I found this discussion pretty funny

[cinap_lenrek]

whats the issue? plan9 has a syscall to create anonymous pipes:

#define	PIPE		21

--
cinap

Re: [9fans] I found this discussion pretty funny

[Charles Forsyth]

[-- Attachment #1: Type: text/plain, Size: 212 bytes --]

but that just turns round and uses the name space (like a few other system
calls).

On 17 October 2012 15:23, <cinap_lenrek@gmx.de> wrote:

> whats the issue? plan9 has a syscall to create anonymous pipes:

[-- Attachment #2: Type: text/html, Size: 459 bytes --]

Re: [9fans] I found this discussion pretty funny

[cinap_lenrek]

yeah, like mount... its files all the way down :)

linux is like windows... everything is a HANDL^Wfiledescriptor

--
cinap

Re: [9fans] I found this discussion pretty funny

[Francisco J Ballesteros]

did you notice the discussion was from this century?
surprising!

On Oct 17, 2012, at 8:28 PM, cinap_lenrek@gmx.de wrote:

> yeah, like mount... its files all the way down :)
>
> linux is like windows... everything is a HANDL^Wfiledescriptor
>
> --
> cinap

Re: [9fans] I found this discussion pretty funny

[tlaronde]

On Wed, Oct 17, 2012 at 06:40:45AM -0700, ron minnich wrote:
> get to the part about why it's so great that pipe is a system call,
> not a device.
>
> This is the sort of back-and-forth that reminds me why I can't quite
> give up on plan 9 ...
>
> http://yarchive.net/comp/linux/everything_is_file.html
>

I like the way the discussion goes. From an outspoken Linus Torvalds
saying "nonsense", to a diving into details that progressively gives
points to some operating system with both "a lot of things are files",
and namespaces---these are related, but not identical.
--
        Thierry Laronde <tlaronde +AT+ polynum +dot+ com>
                      http://www.kergis.com/
Key fingerprint = 0FF7 E906 FBAF FE95 FD89  250D 52B1 AE95 6006 F40C