[9fans] 9P2000 question

[9fans] 9P2000 question

[Fariborz 'Skip' Tavakkolian]

How are additional (ad-hoc or not currently accounted for) file attributes
added to the filesystem? In re-reading the stat/wstat description for 9P2000
there does not seem to be a provision for that. What I think would be useful
is to have additional attributes which may be filesystem specific and that
only certain application may care to know about, etc.

Re: [9fans] 9P2000 question

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 101 bytes --]

If it's just between you and the server, go for it.  The protocol will leave
them alone.

-rob


[-- Attachment #2: Type: message/rfc822, Size: 1729 bytes --]

From: "Fariborz 'Skip' Tavakkolian" <fst@real.com>
To: 9fans@cse.psu.edu
Subject: [9fans] 9P2000 question
Date: Tue, 24 Jul 2001 12:43:13 -0700
Message-ID: <3.0.5.32.20010724124313.013e2c60@mail.real.com>

How are additional (ad-hoc or not currently accounted for) file attributes
added to the filesystem? In re-reading the stat/wstat description for 9P2000
there does not seem to be a provision for that. What I think would be useful
is to have additional attributes which may be filesystem specific and that
only certain application may care to know about, etc.

Re: [9fans] 9P2000 question

[Dan Cross]

In article <20010724201759.073F8199F3@mail.cse.psu.edu> you write:
>If it's just between you and the server, go for it.  The protocol will leave
>them alone.

I think he was refering to the stat[n] field in an Rstat message not
having a place to put non-standard stuff.  I suppose he could just tack
a few bytes onto the end, beyond the muid[n] field, and the protocol
wouldn't care.  The ``normal'' libraries might.

	- Dan C.

Re: [9fans] 9P2000 question

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 44 bytes --]

There are free bits in Qid.type.

-rob


[-- Attachment #2: Type: message/rfc822, Size: 1951 bytes --]

From: Dan Cross <cross@math.psu.edu>
To: 9fans@cse.psu.edu
Cc: 
Subject: Re: [9fans] 9P2000 question
Date: Tue, 24 Jul 2001 17:34:25 -0400 (EDT)
Message-ID: <200107242134.RAA08085@augusta.math.psu.edu>

In article <20010724201759.073F8199F3@mail.cse.psu.edu> you write:
>If it's just between you and the server, go for it.  The protocol will leave
>them alone.

I think he was refering to the stat[n] field in an Rstat message not
having a place to put non-standard stuff.  I suppose he could just tack
a few bytes onto the end, beyond the muid[n] field, and the protocol
wouldn't care.  The ``normal'' libraries might.

	- Dan C.

Re: [9fans] 9P2000 question

[Fariborz 'Skip' Tavakkolian]

>There are free bits in Qid.type.

Certainly Rstat.stat.qid.type does help, but what would one do with things
like title/author/copyright information in a non-hacked way?

Re: [9fans] 9P2000 question

[William K. Josephson]

On Tue, Jul 24, 2001 at 03:34:30PM -0700, Fariborz 'Skip' Tavakkolian wrote:
> >There are free bits in Qid.type.
>
> Certainly Rstat.stat.qid.type does help, but what would one do with things
> like title/author/copyright information in a non-hacked way?

Why does that belong in the filesystem?

  -WJ

Re: [9fans] 9P2000 question

[Fariborz 'Skip' Tavakkolian]

These type of data are intrinsic properties for any copyrighted content.
Depends on the filesystem and on the purpose.

At 06:43 PM 7/24/2001 -0400, William K. Josephson wrote:
>On Tue, Jul 24, 2001 at 03:34:30PM -0700, Fariborz 'Skip' Tavakkolian wrote:
>> >There are free bits in Qid.type.
>>
>> Certainly Rstat.stat.qid.type does help, but what would one do with things
>> like title/author/copyright information in a non-hacked way?
>
>Why does that belong in the filesystem?
>
>  -WJ
>
>

Re: [9fans] 9P2000 question

[William K. Josephson]

On Tue, Jul 24, 2001 at 04:14:59PM -0700, Fariborz 'Skip' Tavakkolian wrote:
> These type of data are intrinsic properties for any copyrighted content.
> Depends on the filesystem and on the purpose.

Sounds like you want a cross reference of some sort, not additional
goo in the file protocol.

 -WJ

Re: [9fans] 9P2000 question

[Dan Cross]

In article <20010724184326.A55109@honk.eecs.harvard.edu> you write:
>> Certainly Rstat.stat.qid.type does help, but what would one do with things
>> like title/author/copyright information in a non-hacked way?
>
>Why does that belong in the filesystem?

Maybe he's making a ``XanaduFS.''

I know....  We need to change 9P2000 to use ASN.1 and run on top
of the OSI protocols.  Then we can have an extensible status message!

	- Dan C.

Re: [9fans] 9P2000 question

[Dan Cross]

In article <20010724215030.AF1F619A01@mail.cse.psu.edu> you write:
>There are free bits in Qid.type.

True, but 5 bits won't even buy you a cup of coffee these days.

	- Dan C.

Re: [9fans] 9P2000 question

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 187 bytes --]

Make a file system that presents an index/annotation file for each
regular file. Doug McIlroy proposed this many years ago, but as far
as I know it's never been implemented.

-rob


[-- Attachment #2: Type: message/rfc822, Size: 1604 bytes --]

From: "Fariborz 'Skip' Tavakkolian" <fst@real.com>
To: 9fans@cse.psu.edu
Subject: Re: [9fans] 9P2000 question
Date: Tue, 24 Jul 2001 15:34:30 -0700
Message-ID: <3.0.5.32.20010724153430.013e2c60@mail.real.com>

>There are free bits in Qid.type.

Certainly Rstat.stat.qid.type does help, but what would one do with things
like title/author/copyright information in a non-hacked way?

Re: [9fans] 9P2000 question

[David Gordon Hogan]

> I know....  We need to change 9P2000 to use ASN.1 and run on top
> of the OSI protocols.  Then we can have an extensible status message!

Wash your mouth out!

Anyway, I though everyone was using XML and CORBA (or DCOM) these days :-)

Re: [9fans] 9P2000 question

[Dan Cross]

In article <20010726031451.78AFE19A3C@mail.cse.psu.edu> you write:
>> I know....  We need to change 9P2000 to use ASN.1 and run on top
>> of the OSI protocols.  Then we can have an extensible status message!
>
>Wash your mouth out!

I'm sorry.

>Anyway, I though everyone was using XML and CORBA (or DCOM) these days :-)

Aha!  Now is the time for you to wash your mouth out.  Everyone knows
that it's XML and J2EE.  Or, if you're from that _other_ camp, XML and
.Net.  Muwhahaa....

	- Dan C.

Re: [9fans] 9P2000 question

[presotto]

I implemented a version of 9P2000 that represents messages in XML and
uses SOAP to marshal.  Unfortunately it was all done on a cocktail napkin
and I ran out of space after the nop message.

Re: [9fans] 9P2000 question

[Fariborz 'Skip' Tavakkolian]

That is funny.

BTW, if you decide to work for the Big-Behemoth-Software-Company across the
lake from here, they'll take this seriously. Knowing something about the
culture
there, they'll stop reading before the last sentence and initiate a project
immediately to "embrace and extend" 9P2000.

This has gone off the original subject. I get the joke. The most constructive
suggestions so far are from rob (thanks).

At 09:06 AM 7/26/2001 -0400, presotto@plan9.bell-labs.com wrote:
>I implemented a version of 9P2000 that represents messages in XML and
>uses SOAP to marshal.  Unfortunately it was all done on a cocktail napkin
>and I ran out of space after the nop message.
>
>

Re: [9fans] 9P2000 question

[Dan Cross]

In article <20010726130635.8C63819A10@mail.cse.psu.edu> you write:
>I implemented a version of 9P2000 that represents messages in XML and
>uses SOAP to marshal.  Unfortunately it was all done on a cocktail napkin
>and I ran out of space after the nop message.

It was probably faster than most SOAP implementations.

	- Dan C.