[9fans] [OT] domain socket support in open(2) on unix

[9fans] [OT] domain socket support in open(2) on unix

[William Ahern]

i figured this was OT, but this might be the only group who can _hear_ what
i'm asking.

could you guys comment on the efficacy of adding client-side unix
domain socket support to the unix open(2) library and/or system call?

over on comp.unix.programmer i'm getting the obvious reponses, "domain
sockets were meant as an adjunt to the more general network socket
interface". fine. but my point is (well, and question as to the practicality
of) that domain sockets can be appropriated to provide user-land support for
_device_ style file objects.

i'm just curious as to what your views are. useful? no utility because
the standard shells won't reflect this capability?

tia,

Bill

Re: [9fans] [OT] domain socket support in open(2) on unix

[boyd, rounin]

everything is a file ...

Re: [9fans] [OT] domain socket support in open(2) on unix

[ron minnich]

hey, interesting thing. We had a guy from Monash here -- Abramson. Very
smart guy. He gave a good talk and described his work in grids.

Guess what it turns out grid programmers want to use? Not
GLOBUS_read_from_some_kinda_magic_globus_file_handle_on_odd_days.
Nosirree.

What they want, it turns out, is this:
read
write

That's about it.

Oh, wait, forgot the hard ones:
open
close

On the open, they want it to look like a file name.

Pretty wild, huh? Is there a place in here for Plan 9, possibly? :-)

ron

Re: [9fans] [OT] domain socket support in open(2) on unix

[boyd, rounin]

> Oh, wait, forgot the hard ones:
> open
> close

don't tease me ...

Re: [9fans] [OT] domain socket support in open(2) on unix

[Russ Cox]

> Oh, wait, forgot the hard ones:
> open
> close

They're not that bad.  We can split the work.
You do open, I'll do close.

Russ

Re: [9fans] [OT] domain socket support in open(2) on unix

[boyd, rounin]

> You do open, I'll do close.

can we do it in texas?

Re: [9fans] [OT] domain socket support in open(2) on unix

[Dan Cross]

Hey, I'll be in Texas.  Let me get in on something.  I know, I'll do
name caching....

	- Dan C.

Re: [9fans] [OT] domain socket support in open(2) on unix

[northern snowfall]

>
>
>They're not that bad.  We can split the work.
>You do open, I'll do close.
>
No way, man. I heard it takes at *least* five
hackers working round-the-clock to open a
file descriptor.
At Bell Labs this might be optimized to four.
I'm not sure, I dont' have the statistics...

>

Re: [9fans] [OT] domain socket support in open(2) on unix

[andrey mirtchovski]

On Tue, 20 May 2003, ron minnich wrote:

> Pretty wild, huh? Is there a place in here for Plan 9, possibly? :-)
>

absolutely not: plan 9 is a non-unix system, the license sucks and the
community is closed and elitist.

notice how none of those have anything to do with the technical merit of the
operating system?

andrey

[9fans] Re: [Way OT] domain socket support in open(2) on unix

[Jim Choate]

On Tue, 20 May 2003, andrey mirtchovski wrote:

> On Tue, 20 May 2003, ron minnich wrote:
>
> > Pretty wild, huh? Is there a place in here for Plan 9, possibly? :-)
> >
>
> absolutely not: plan 9 is a non-unix system, the license sucks and the
> community is closed and elitist.
>
> notice how none of those have anything to do with the technical merit of the
> operating system?

Man cannot live by bread alone.


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] [OT] domain socket support in open(2) on unix

[okamoto]

> absolutely not: plan 9 is a non-unix system, the license sucks and the
> community is closed and elitist.
>
> notice how none of those have anything to do with the technical merit of the
> operating system?

They don't need to pay too much attention to the present status as is,
which may lead deep innovation in a sense of technology.
Am I wrong? ?

Kenji

Re: [9fans] [OT] domain socket support in open(2) on unix

[andrey mirtchovski]

On Wed, 21 May 2003 okamoto@granite.cias.osakafu-u.ac.jp wrote:

> They don't need to pay too much attention to the present status as is,
> which may lead deep innovation in a sense of technology.
> Am I wrong? ?
>

I have no idea, really.. I am just reiterating what I've heard over the
past three years. If you're suggesting that Plan 9 may have a higher
future return in terms of research and development than UNIX, then I
hope you are right...

In my opinion, it's just history repeating -- "buying IBM never fired
anybody" (wasn't that what they used to say a few decades ago?).

s/IBM/Lunix/, s/buying/using/.

andrey

Re: [9fans] [OT] domain socket support in open(2) on unix

[ron minnich]

On Tue, 20 May 2003, andrey mirtchovski wrote:

> In my opinion, it's just history repeating -- "buying IBM never fired
> anybody" (wasn't that what they used to say a few decades ago?).

don't give up. The OSes we curse now (Unix and  Linux) are what we wanted
to have 10 and 20 years ago, when we were cursing miserable vendor Oses.

So, in the best case, you can look forward to the day when we're all using
Plan 9 and cursing our luck for being stuck with it. Or you're using it
and liking it and all these college students who look 6 years old are
telling you (respectfully, of course): "Dr. Mirtchovski, you're so old
school. Can I bring you your walker? Are your teeth in today?"

That's called 'progress'.

Feel better?

ron

Re: [9fans] [OT] domain socket support in open(2) on unix

[Geoff Collyer]

`Nobody ever got fired for buying IBM' was the old saying.  I think
the modern equivalent of IBM in that saying is Wintel, with Linux
gaining fast in the server market.

Re: [9fans] [OT] domain socket support in open(2) on unix

[okamoto]

> past three years. If you're suggesting that Plan 9 may have a higher
> future return in terms of research and development than UNIX, then I
> hope you are right...

I don't think that Plan 9 may have a higher future in the sense of
marketting.  In that sense, Windows have spreaded and controled
over all the people in this real world.   No other operating system
is there such that.   Then, are you satisfied with this situation?
I have confidence that you ain't, because you are here.   There may be
still people who want open and clear operating system for their purpose...
Linux may be the Windows in the Unices world. ?

Kenji

[9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[Jim Choate]

On Wed, 21 May 2003 okamoto@granite.cias.osakafu-u.ac.jp wrote:

> still people who want open and clear operating system for their purpose...
> Linux may be the Windows in the Unices world. ?

The one overriding character of every comment in this thread is
monotonicity. It's one or the other but not both.

How sad. You'd think people would realize that there is -no one answer-.

Go re-read your economic and political theory books. Pay attention to
jewels like Federalist #10 or Hayek.

The problem is fundamentaly one of biology, not technology.


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[Lucio De Re]

On Tue, May 20, 2003 at 10:47:36PM -0500, Jim Choate wrote:
>
> The problem is fundamentaly one of biology, not technology.
>
Or literacy, but I digress.

This community isn't going to solve a biology problem.  We can
address technological ones, however, give or take occasional pokes
from the sidelines.

I vote that we stick to what we do best.  My forays into biological
manipulation I limit to reproductive activities.  I'm sure they
don't interest anyone else on this list and even if they did, I
believe in being discrete :-)

++L

Re: [9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[boyd, rounin]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 312 bytes --]

"Lucio De Re" <lucio@proxima.alt.za> a �crit dans le message de
news:20030521064709.U7647@cackle.proxima.alt.za...
> I vote that we stick to what we do best.

John Shaft [Samuel L. Jackson to Peoples Hernandez:]:

    You're a 2 bit, 3 block, drug dealer ... stick to what you know

you got my vote

Re: [9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[Jim Choate]

On Wed, 21 May 2003, Lucio De Re wrote:

> On Tue, May 20, 2003 at 10:47:36PM -0500, Jim Choate wrote:
> >
> > The problem is fundamentaly one of biology, not technology.
> >

> Or literacy, but I digress.

Another biology problem I see.

> This community isn't going to solve a biology problem.  We can
> address technological ones,

False distinction on your part. Without biology there is no technology.

Your view however is an example of a biological problem with this
community.


 --
    ____________________________________________________________________

      We are all interested in the future for that is where you and I
      are going to spend the rest of our lives.

                              Criswell, "Plan 9 from Outer Space"

      ravage@ssz.com                            jchoate@open-forge.org
      www.ssz.com                               www.open-forge.org
    --------------------------------------------------------------------

Re: [9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[Lucio De Re]

On Wed, May 21, 2003 at 06:27:57AM -0500, Jim Choate wrote:
>
> Your view however is an example of a biological problem with this
> community.
>
And your solution, Dr Mengele, is?

++L

Re: [9fans] The Future of OS's & P9 (was: Re: [OT] domain socket support in open(2) on unix)

[boyd, rounin]

> And your solution, Dr Mengele, is?

thalium

Re: [9fans] [OT] domain socket support in open(2) on unix

[Geoff Collyer]

It seems obvious that it should be possible to open existing
Unix-domain sockets like any other file.  They appear in the
file-system namespace, after all.  Other systems call them named pipes
or mounted streams.  It should be a trivial change to creat to create
Unix-domain sockets.

Of course, the whole range of socket-related system calls could be
replaced with /net, dial and cs, and this would eliminate arcane and
often buggy socket code from ordinary programs, allowing them to adapt
without recompilation to, for example, IP V6.

Re: [9fans] [OT] domain socket support in open(2) on unix

[Russ Cox]

> i'm just curious as to what your views are. useful? no utility because
> the standard shells won't reflect this capability?

if done right, the standard shells shouldn't need to know.
open should just work.