[Sussex] Improving on UNIX

Geoff Teale Geoff.Teale at claybrook.co.uk
Fri Mar 14 08:48:00 UTC 2003


Morning...

What a wonderful morning it is.. shame I'm at work :(

I was just browsing around over at http://www.kuro5hin.org  - always a good
place for some random wittering.  I came across an article proclaiming
itself to be a discussion of how to make an Object-Orientated GUI layer for
UNIX that embraced and extended the founding principles of UNIX.
Unfortunately this article turned out to be a little bit of a rant by a
Mac-Ophile about how good the old MacOS was and how Mac OS X, KDE and Gnome
wanted to be Windows too much.

It was an interesting idea, it's just the article wasn't that well executed.
I wonder if it is possible to provide the power of UNIX without just serving
up a command line - answers on a postcard please ;)

Two interesting comments followed the article.  

The first bemoaned the death of BeOS - now here I have to agree - like UNIX,
the foundation of BeOS was the filesystem and the transparancy of the system
through a filesystem interface.  BeFS and the interface to it (Tracker) were
the UNIX filesystem evolved to a near perfect state.  BeFS was (is) a
64-bit, fully journaled filesystem with dynamic attribute support - it was
way ahead of the game in it's day and has, to my knowledge, not been matched
for useability yet.  The important thing was that it was designed to perform
like a database - dynamic attribute support meant you could store as much
meta-data around a file as you liked and then you could run queries against
the filesystem.  All of this functionality was made availble through the
Tracker (the BeOS filesystem browser / desktop).  The best thing was that
with a little imagination you could build whole applications in Tracker -
the default BeOS mailing system used a Mail Daemon to fetch and send into
folder - attributes of each file (eMail) where displayed and functions
(open, reply, forward) as context sensitive menus.  The Address book was
just another directory, each entry was a zero-length file with a number of
attributes, just as with E-mails the files were identified as Contacts by
their mime-type and handled in the way the system had specified for Contact
files.  

All of this functionality was there because of extensions to the filesystem
and the OS built upon it - but, of course the filesystem itself was just a
standard UNIX filesystem to application code that didn't support the
extensions.

The big problem of course was the BeOS wasn't of much commercial value - it
was wonderful in the context of Windows 95 - it had the same (zero) security
model and you could do 90% of everything a home user wants to do, but you
couldn't put it in a commercial environment - it could mount NFS shares
(insecurely) and talk AppleTalk but SMB interaconnectivity ("World 'O'
Networking") was borked big time and the whole damn thing was single user
and had _terrible_ network performance (although I know a guy who still runs
his business network (30 machines) with a BeOS machine running Communigate
Pro).

...and thus BeOS was doomed.  But the thing is there are people out there
trying to build an OpenSource version of BeOS. The first group of people are
trying to rebuild the whole thing from scratch - good luck to them I say,
that platform was great, but it was always flawed in the context of the
real-world and frankly their simply isn't enough software available for it.
There is however at least one group who are trying to rebuild BeOS on
Linux/XFree86.  Now when I first heard this I thought it was a bad idea, but
thinking about it now I'm kind of hankering to take a look.  If that project
could deliver BeFS and the BeOS desktop (with tracker) intact _and_ provide
100% compatibility with true LINUX systems (can't see why not) then
effectively it just becomes a choice of filesystem and GUI environment and
then we'd be cooking on gas - because no matter how good Gnome, KDE, etc.
get, they are just a top end product - the BeOS GUI was bound implicitly to
it's filesystem  and it gave a working environment that no system before or
since has ever matched.  I'm going to do some more research - see if I can't
get some of the stuff out there working on a machine someplace ;)

The second interesting comment came about because in the original article
the following throw-away line was proffered:

"You we really hate out users that much?"

The simple, one word answer that came back was:

"Yup"

... sometimes life is beautiful.. :)


-- 
geoff.teale at claybrook.co.uk
tealeg at member.fsf.org

"make music like mercy that gives what it is and has nothing to prove"
 - Ani DiFranco "Up Up Up Up Up Up"


The above information is confidential to the addressee and may be privileged.  Unauthorised access and use is prohibited.
 
Internet communications are not secure and therefore this Company does not accept legal responsibility for the contents of this message.
 
If you are not the intended recipient, any disclosure, copying, distribution or any action taken or omitted to be taken in reliance on it, is prohibited and may be unlawful.
 
Claybrook Computing Limited is a subsidiary of Claybrook Computing (Holdings) Limited
Registered Office: Abbey House. 282 Farnborough Road, Farnborough, Hampshire GU14 7NJ
Registered in England and Wales No 1287205
 
A Hogg Robinson plc company





More information about the Sussex mailing list