Re: vms versus solaris
From: Bill Todd (billtodd_at_metrocast.net)
Date: 01/17/05
- Next message: Bill Gunshannon: "Re: vms versus solaris"
- Previous message: Tom Linden: "Re: vms versus solaris"
- In reply to: JF Mezei: "Re: vms versus solaris"
- Next in thread: prep_at_prep.synonet.com: "Re: vms versus solaris"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Mon, 17 Jan 2005 14:50:44 -0500
JF Mezei wrote:
...
> Actually, VMS and Apple have an untapped edge which may be realised in a
> couple of years.
I wouldn't count on that.
>
> Consider what Microsoft wants to do with its file system (causing many
> years of delays in its introduction).
What Microsoft wants to do is create a data environment sufficiently
complex and integrated that other systems can't effectively duplicate
it, thus strengthening its stranglehold on its customers. Fortunately,
they've discovered that doing that anywhere nearly right (i.e., creating
something actually useful) is quite difficult, and (perhaps
surprisingly) haven't succumbed to the temptation just to throw
something half-baked over the wall in the interim (perhaps in part
because they remember how poorly their first such attempt - OLE
structured storage - fared in this respect, and for that matter how
little NTFS's advanced features - primarily, attributes/streams - have
been used).
>
> The presence of an RMS layer on VMS allows the VMS engineers to add a
> lot of file features transpararently to existing applications.
Only to well-behaved, forward-looking applications - e.g., those that
won't get upset if they see an unknown XAB type when opening a file.
A Unix environment could be extended as well: their interfaces may not
have been designed to be, but that only means that the extension syntax
would lack elegance, not that it would be anything like impossible.
...
> Lets say VMS engineers were tasked to add new RMS fields such as a
> "comments", "source URL" and "keywords" fields. This could be done by
> just adding fields to the FAB structure, probably an XAB. (think of the
> use of comments when storing images for instance).
>
> On the simple file systems such as Unix and Windows, this is not
> possible without major rewrite of the file system. On VMS, the hooks are
> already in, it is just a question of adding the relevant functionality.
Welllll - not exactly.
It would certainly be possible to enhance relative and indexed files to
contain such data in their prologues, but only by creating new prologue
version levels which older VMS systems couldn't handle - not exactly the
kind of friendly bi-directional compatibility that one might wish for.
But there's no prologue in sequential files where such data could be
maintained: the file system itself (not RMS) would need to be extended,
perhaps by creating new areas in the index file header (which is kind of
scarce and valuable real estate for such use, though NTFS-like
multi-stream facilities might be feasible and may even currently exist
to some degree - I seem to remember something along those lines a while
back, and of course this would not be all that difficult for Unix-style
file systems as well, using extended filespec syntax to access them just
as NTFS does rather than call-interface extensions per se).
And of course even with the file system so extended the only real
advantage RMS would be bringing to the effort would be the ability to
extend the RMS parameter-block interface somewhat less kludgily than the
Unix call interface (if they didn't elect to go the extended filespec
syntax route). Given that the vast majority of applications access
files through higher-level language interfaces (which would have to be
extended in either environment before the added functionality became
easily accessible to such applications - yet again, assuming that it
wasn't accessed via filespec extensions), that advantage seems
relatively marginal.
- bill
- Next message: Bill Gunshannon: "Re: vms versus solaris"
- Previous message: Tom Linden: "Re: vms versus solaris"
- In reply to: JF Mezei: "Re: vms versus solaris"
- Next in thread: prep_at_prep.synonet.com: "Re: vms versus solaris"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
