> Mumble. I'm thinking about folding namer paths into the main open() code
> path. Currently they're distinct, and it's causing most of your confusion.
>
> The namer path should disk/wd. This maps a path to a port number. When you
> mount that port number in your namespace, the server at that port number
> looks like a directory tree. So disk/wd:wd0_dos0 is shorthand,
> conceptually, for "mount disk/wd on /FOO, and then open /FOO/wd0_dos0". I
> say conceptually, because it doesn't really join your filesystem namespace.
[Andrew's mounting explanation deleted]
That also cleared up some confusion I had about the whole disk/wd:??
bussiness. Into the FAQ it goes :)
However, it reinforces the previous thoughts I had. I assume (perhaps
incorrectly) that when you said you were thinking of folding namer paths
into the main open() code path that "disk/wd" would essentally be part of
the man open() code path, thus you could directly mount "disk/wd/wd0_dos0"
somewhere? While I think the shorthand and concept behind the
disk/wd:wd0_disk0 is slightly confusing (espectially before a proper
explanation), I think it would make more sense to keep the namer paths
separate, but allow them to work slightly more like normal paths,
in that "disk/wd/wd0_dos0 would mean to open "wd0_dos0" from the
server available in the namer path "disk/wd". Perhaps these things are nearly
mutually exclusive, or perhaps I'm still confused about how namer paths
are distinct from the main pathspace.
After having said that, I would like to test if I understand the idea
behind the current layout. If I were to create a second "namer" which would
make itself available in the "real" first namer, and then run wd and
have itself registered in the second namer, would it then be logical to
use "namer2:disk/wd:wd0_dos0" to refer to the partition?
If this concept is correct, then I understand the logic behind things
and I really think that just a good expanation is the real solution to the
confusion.
-- jeske_at_uiuc.edu + David Jeske(N9LCA)<A HREF="http://www.cen.uiuc.edu/~jeske/"> NeXTMail accepted + CompEng Student/NeXT Programmer/Call Gtalk at (708)998-0008 User of Linux/NEXT/DOS/WIN/OS.2/VSTa (all coexisting on one system) </A> -- jeske_at_uiuc.edu + David Jeske(N9LCA)<A HREF="http://www.cen.uiuc.edu/~jeske/"> NeXTMail accepted + CompEng Student/NeXT Programmer/Call Gtalk at (708)998-0008 User of Linux/NEXT/DOS/WIN/OS.2/VSTa (all coexisting on one system) </A> FromFrom vandys@puli.cisco.com Mon Jun 19 18:31:05 1995 Received: from hubbub.cisco.com (hubbub.cisco.com [198.92.30.32]) by amri.cisco.com (8.3/8.3) with ESMTP id SAA00139; Mon, 19 Jun 1995 18:31:04 -0700 Received: from w4sun5.ccl.itri.org.tw (w4sun5.ccl.itri.org.tw [140.96.111.5]) by hubbub.cisco.com (8.6.10/CISCO.GATE.1.1) with SMTP id CAA07517 for <vsta@cisco.com>; Mon, 19 Jun 199 vandys@puli.cisco.com Mon Jun 19 18:31:12 1995 Received: from hubbub.cisco.com (hubbub.cisco.com [198.92.30.32]) by amri.cisco.com (8.3/8.3) with ESMTP id SAA00141; Mon, 19 Jun 1995 18:31:06 -0700 Received: from disperse.demon.co.uk (disperse.demon.co.uk [158.152.1.77]) by hubbub.cisco.com (8.6.10/CISCO.GATE.1.1) with SMTP id BAA06138 for <vsta@cisco.com>; Mon, 19 Jun 1995 01:37:05 -0700 Received: from punt.demon.co.uk by disperse.demon.co.uk id ab12377; 19 Jun 95 9:28 +0100 Received: from humbug.demon.co.uk by punt.de5 02:36:30 -0700 Received: from cclw400.ccl.itri.org.tw by w4sun5.ccl.itri.org.tw (4.1/SMI-4.1) id AA19379; Mon, 19 Jun 95 17:50:29 CST Received: from CCLW400/SpoolDir by cclw400.ccl.itri.org.tw (Mercury 1.13); Mon, 19 Jun 95 17:30:48 gmt+800 Received: from SpoolDir by CCLW400 (Mercury 1.13); Mon, 19 Jun 95 17:30:40 gmt+800 From: "James Lee" <JLEE@CCLW400.CCL.ITRI.ORG.TW> Organization: ITRI/CCL/W400 To: vsta@cisco.com Date: Mon, 19 Jun 1995 17:30:37 GMT+800 Subject: vsta info Priority: normal X-Mailer: Pegasus Maimon.co.uk id aa15692; 19 Jun 95 9:28 +0100 Received: by humbug.demon.co.uk (Linux Smail3.1.29.1 #3) id m0sNbun-00036rC; Mon, 19 Jun 95 09:10 BST Message-Id: <m0sNbun-00036rC@humbug.demon.co.uk> From: Dave Hudson <dave@humbug.demon.co.uk> Subject: Re: VSTa mounting and paths To: jeske@uiuc.edu Date: Mon, 19 Jun 1995 09:10:36 +0100 (BST) Cc: VSTa mailing list <vsta@cisco.com> In-Reply-To: <199506161956.OAA16620@fiction.isdn.uiuc.edu> from "David Jeske" at Jun 16, 95 02:56:43 pm X-Mailer: ELM [version 2.4 PL23] MIME-Versiol/Windows (v1.22) Message-Id: <3A4BFC70BD0@cclw400.ccl.itri.org.tw> Please email me more info about vsta. Thanks. James Lee Computer & Communications Lab Industrial Technology Research Institute phone: 886-35-78-1390x303 fax: 886-35-78-1398 jlee@ccl.itri.org.tw n: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 2536 Hi, David Jeske wrote: > > > Mumble. I'm thinking about folding namer paths into the main open() code > > path. Currently they're distinct, and it's causing most of your confusion. > > > > The namer path should disk/wd. This maps a path to a port number. When you > > mount that port number in your namespace, the server at that port number > > looks like a directory tree. So disk/wd:wd0_dos0 is shorthand, > > conceptually, for "mount disk/wd on /FOO, and then open /FOO/wd0_dos0". I > > say conceptually, because it doesn't really join your filesystem namespace. > > However, it reinforces the previous thoughts I had. I assume (perhaps > incorrectly) that when you said you were thinking of folding namer paths > into the main open() code path that "disk/wd" would essentally be part of > the man open() code path, thus you could directly mount "disk/wd/wd0_dos0" > somewhere? While I think the shorthand and concept behind the > disk/wd:wd0_disk0 is slightly confusing (espectially before a proper > explanation), I think it would make more sense to keep the namer paths > separate, but allow them to work slightly more like normal paths, > in that "disk/wd/wd0_dos0 would mean to open "wd0_dos0" from the > server available in the namer path "disk/wd". Perhaps these things are nearly > mutually exclusive, or perhaps I'm still confused about how namer paths > are distinct from the main pathspace. I think the confusion comes because it's difficult to differentiate a server path from a normal path. The obvious solution that springs to mind is to prefix server paths with a magic character. We already have "@" (if I remember correctly) for handling environment path expansions. Plan 9 devices use a prefix hash symbol "#" (I believe this is known as a pound symbol in the US (or at least that's what the Plan 9 docs call it), but that confuses me as a pound symbol is the currency one (looks like a squiggly L) over here :-)). Perhaps the solution is to patch libc to allow something similar, thus we have: #disk/wd:wd0_dos0 as a server/path, and the more conventional /dev/wd/wd0_dos0 for when #disk/wd has been mounted at /dev. This might mean a reasonable amount of extra work to patch some of the utils (actually removing quite a bit of code I suspect), but would give us a pretty consistent view. If nothing else, adopting something like this gives us a consistent way to name files within VSTa (it's implicit whether we have a server reference or a "mounted" pathname). Regards, DaveReceived on Sun Jun 18 07:22:22 1995
This archive was generated by hypermail 2.1.8 : Thu Sep 22 2005 - 15:12:27 PDT