> > : matomira
> : mapleson
> > Well, I have always thought SGI should make PCs, given their presence
> > in all markets (including game consoles) except this. ...
>
> This contradicts your later anti-MS comments.
PC != M$
> The Cray buy was a very good move IMO. SGI should always maintain a tech
> base that's skilled enough to design stuff like the HUB ASIC (have you
> read the report on that device? If not, do so).
Well, I also like SGI owning Cray, but maybe mostly for `coolness' reasons.
What was shortsighted was concentrating on a big iron business w/o starting
to move into a big volume market. I don't know how many patents (besides
maybe CLI?) this brought in, otherwise it would have been possible just
to hire the people after Cray sank. 1st year of Challenge introduction
grabbed 30% of the supercomputer market (OK. low-end class. So what?)
Merging of the disparate technologies takes at least 5 years. Good that they
sold to Sun the SPARC-based line.
> > I think the `NT' thing is consistent with the traditional SGI strategy,
> > that is, make great machines running bad software (C/C++/Unix) ...
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>What a wierd comment. Funny how a post can so flatly contradict one's
>own experience. I've been using IRIX for years and I still think it's
>the best OS I've used yet; it's stability and user-interface are delightful.
Of course IRIX is the best Unix around (which BTW, marketing never knew
how to sell, I mean, where are the ads explaining how cool N32 is?).
But it's just a C/C++ OS. Light-years ago from the quality attainable
by Lisp machines.
> Btw, C and C++ are programming languages, not 'software'. Or were
> you criticising the MIPS Pro Compilers?
Criticising fascist C++-oriented environment, and general lack of lisp
literacy in the developer community. SGI gives you Pascal, Ada, C, C++
compilers but they just don't care about lisp (while Sun did, at least a
bit). And lisp vendors are too concentrated on the traditional AI people. So
when you want to do graphics with Lisp and SGI tools is a big hassle.
[SIDENOTE: Symbolics Graphics Division software is now Nichimen's]
> > ... because that's
> > what everybody wants. ...
>
> Flatly wrong. You may have meant 'most people', or 'most PC users', but
> that's not what you said. What did you actually mean? Please write what
> you really mean, otherwise one has no way of knowing your real opinions.
Yes. I forgot to say `almost'. I for one, want _The Real Thing_.
> > That said, I remember the old slogan "SGI - Inspired by Vision". I think
> > the new one should be "SGI - NO Vision (at all)". They should not be
> > cannibalizing IRIX. Commitment to quality should rule NT out. ...
> What they should or should not be doing is irrelevant. What matters first
> and foremost is market realities, and the basic market reality of today
> is that many companies want NT support, whether or not they're aware
> of its flaws.
NT support is one thing. Destroying the company identity is another.
> And moves like Fahrenheit are definitely sensible. Time to put a stop
> to the bloated PC API mess.
But you don't need to build a shrine to M$. And if so many developers
are taking up OpenGL. Do you really need M$? Ah, kernel stuff..
> > NT on SGI would have been a good option to offer. You do things
> > right when they should, or you just don't.
>
> How can you judge when the VPC isn't even out yet? You don't know what
> SGI will be offering the NT world.
I meant NT on _real_ SGI, i.e. MIPS. circa 1994.
> Well, technically it already is offering things since Maya is available for
> NT.
So? You can run that on Intergraph..
> > PC != NT. ...
>
> For many, there is little difference. That's the measure of the grip
> MS has on the PC world.
>
> > ... As said a thousand times, PC with Linux is OK. ...
>
> OK is not good enough for professional businesses and services. I watched
> a friend try to install Linux recently - what a mess!
What about those companies doing Linux PC rollouts in the 500's? What
about the U.S. Postal Service? What about 300K$ Linux clusters?
I had no problem with RedHat 5.0 (for dummies) and S.u.S.E. 5.3 (a bit more
sophisticated). RedHat is better to single-click install access to an SMB
printer though.
> Linux is fine, but if I owned a business I definitely wouldn't rely on
> it just yet.
What about 1 to 2 years?
> > ... www.sgi.com is
> > no place for ANY Microsoft references at all.
>
> You know this statement is illogical, so why make it?
OK. Maybe some reference in the line `Yeah. We support interoperability
with that and all that other stuff too'. No references that make M$ the center
of the world. SGI stock went down the day NT machines were announced.
> > And NT is dead meat anyway.
>
> This may convey your emotional feelings about NT, but it says nothing
> of value wrt NT's real future.
If you want something to happen first you have to believe it. I would have
not believed it 6 months ago. Now I do.
> I'd far rather read how people feel NT is going to affect SGI's future
> direction, than their opinions as to NT's similarity to decomposed
> animal protein. There's an NT4 box in front of me.... ummm.... noooo,
After not touching a PC for 8 years, and being NT
a `Digital product', I thought it would be OK. Wrong. 1 month only
to discover that it actually meant NewTrash, and later, 1 week
only to find out that VC++ stands for `Vicious C++'.
> [IA64]
> People who worked on those emailed me explaining the situation as outlined
> in my previous paragraph. I believe them. What right have I not to?
sgi.bad-attitude. Oops.
> In the end, it makes no sense at all for us to judge SGI's plans
> when we're not privvy to the detailed information. If you believe such
> ill-informed thinking is a sound methodology, ie. making judgements without
> facts, then why not apply it everywhere?
SGI. Sorry history of misguided top-level mgmnt after Jim Clark left.
I am not sure which is worst: Apple or SGI's.
--
Fernando D. Mato Mira
Real-Time SW Eng & Networking
CSEM
CH-2007 Neuchatel
Switzerland
-----------== Posted via Deja News, The Discussion Network ==----------
http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own
In article <············@nnrp1.dejanews.com>, <········@acm.org> wrote:
->> And moves like Fahrenheit are definitely sensible. Time to put a stop
->> to the bloated PC API mess.
->
->But you don't need to build a shrine to M$. And if so many developers
->are taking up OpenGL. Do you really need M$? Ah, kernel stuff..
I find it ever-so-ironic that SGI is going to a 100% Microsoft
solution, while game (and other) developers are demanding SGI
technology (OpenGL) in Microsoft products. Now there is this strange
FreeBSD/Linux explosion going on and SGI is moving _away_ from it
all. Can we delete the comp.sys.sgi newsgroup yet?
->> Linux is fine, but if I owned a business I definitely wouldn't rely on
->> it just yet.
->
->What about 1 to 2 years?
Hell, what about *now*. Our entire department runs on Linux machines
(about 12 of them), and this is a college of pharmacy. Many people
still remain in denial though (which is why FreeBSD is out there, to
give them an 'out' for not running Linux 'crap' :)
->> In the end, it makes no sense at all for us to judge SGI's plans
->> when we're not privvy to the detailed information. If you believe such
->> ill-informed thinking is a sound methodology, ie. making judgements without
->> facts, then why not apply it everywhere?
->
->SGI. Sorry history of misguided top-level mgmnt after Jim Clark left.
->I am not sure which is worst: Apple or SGI's.
Heh, at least Apple is moving *toward* UNIX.
--
The wheel is turning but the hamster is dead.
Craig Kelley -- ········@isu.edu
http://www.isu.edu/~kellcrai finger ···@inconnu.isu.edu for PGP block
Craig Kelley wrote:
> ->> Linux is fine, but if I owned a business I definitely wouldn't rely on
> ->> it just yet.
> ->
> ->What about 1 to 2 years?
>
> Hell, what about *now*. Our entire department runs on Linux machines
> (about 12 of them), and this is a college of pharmacy. Many people
> still remain in denial though (which is why FreeBSD is out there, to
> give them an 'out' for not running Linux 'crap' :)
When Linux has solid, hardware-accelerated OpenGL, it will make an
interesting alternative to SGI, HP, DEC, Sun, IBM, and NT for
low-end/mid-range 3D graphics. For all the crap that NT is, at least it
has good OpenGL support.
--
Benjamin Z. Goldsteen
Crystallography Program
Oklahoma Medical Research Foundation
405 271 8954 / 405 271 7953 (FAX)
In article <·················@zeta.omrf.ouhsc.edu>,
Benjamin Z. Goldsteen <··················@ouhsc.edu> wrote:
->When Linux has solid, hardware-accelerated OpenGL, it will make an
->interesting alternative to SGI, HP, DEC, Sun, IBM, and NT for
->low-end/mid-range 3D graphics. For all the crap that NT is, at least it
->has good OpenGL support.
Depending on which 3D card you purchase, so does Linux (although we
can't call Mesa OpenGL, now can we....). There was an entire issue of
the Linux Journal devoted to 3D/graphics a couple of months ago.
We are replacing one of our RS/6000 machines with a Linux box which
costs 1/10th as much and has much better 3D performance. It will be
used to continue research into the effect nuclear material has on
brain tumors (it uses OpenGL to render tumor locations in the brain
based on 2D slices).
--
The wheel is turning but the hamster is dead.
Craig Kelley -- ········@isu.edu
http://www.isu.edu/~kellcrai finger ···@inconnu.isu.edu for PGP block
On Thu, 05 Nov 1998 15:03:56 -0600, "Benjamin Z. Goldsteen"
<···@zeta.omrf.ouhsc.edu> claimed or asked:
% When Linux has solid, hardware-accelerated OpenGL, it will make an
% interesting alternative to SGI, HP, DEC, Sun, IBM, and NT for
% low-end/mid-range 3D graphics. For all the crap that NT is, at least it
% has good OpenGL support.
This is because Microsoft could license and subvert OpenGL. Open.
What a dumb name for something that isn't. That is why I have MESA.
If it is at all feasible, I might port Mesa to CL.
Linux will get better hardware support over time. It has credibility
now.
Lisp, for some strange reason, does not. The more I learn about the
language, the more I wonder why it isn't in more general use. Oh
well. There is a simple solution. Go Linux (or FreeBSD or whatever)
when possible as an alternative to NT (W2K, the new millennium bug).
Program large systems in Lisp.
For CS courses, mention that Lisp isn't just for AI. If it can do the
hard stuff, it stands to reason it can do the easy stuff too and
everything in between.
I expect the high end graphics machines of the near future to be
ordinary desk top PCs (not necessarily Intel based) performing in
parallel on a network. Lisp makes for a low bandwidth means of
passing a lot of information. Unix is good at networking. It is
really a good combination, even if Unix is far from perfect.
--
David Steuber (ver 1.31.2a)
http://www.david-steuber.com
To reply by e-mail, replace trashcan with david.
"Ignore reality there's nothing you can do about it..."
-- Natalie Imbruglia "Don't you think?"
In article <··················@news.newsguy.com>,
········@david-steuber.com wrote:
> Linux will get better hardware support over time.
A lot of stuff is getting better hardware support over time.
> The more I learn about the
> language, the more I wonder why it isn't in more general use.
Maybe it is not that easy? You need some time and dedication
to reach the state of competent Lisp programming. Some people
have a really hard time understanding CL. These need
to work a little harder. ;-)
Our company is more on the small side, but we are currently trying to
to do more programming projects. Our inhouse Internet tools are
mostly written in CL/Scheme (more moving to CL). Some customer
projects have been written in Lisp and we are currently
preparing one or two new projects using Lisp, also trying to introduce
more efficient tools.
My long term strategy. O.k., let's waste a few minutes ;-) :
;;; ------------------------------------------------------------
;;; Author: Rainer Joswig, ······@lavielle.com
;;; 7.Nov.1998
;;; ------------------------------------------------------------
;;; Setup
(in-package :cl-user)
(assert (member :cl-http *features*)
(*features*)
"CL-HTTP has to be loaded!")
;;; ------------------------------------------------------------
;;; What is a project?
(defclass project ()
((name :accessor project-name
:initarg :name
:type string)
(time-frame :accessor project-time-frame
:initarg :time-frame
:initform nil)
(sub-projects :accessor project-sub-projects
:initarg :sub-projects
:initform nil)
(reason :accessor project-reason
:initarg :reason)
(state :accessor project-state
:initarg :state)
(steps :accessor project-steps
:initarg :steps)
(parts :accessor project-parts
:initarg :parts)
(description :accessor project-description
:initarg :description)
(url :accessor project-url
:initarg :url)))
(defmacro defproject (symbol (name &key time-frame url) &body body)
`(setf ,symbol
(make-instance 'project
:name ,name
:time-frame ,time-frame
:url ,url
:sub-projects (loop for (key name . args) in ',body
collect (list key (apply #'make-instance 'project
:name name
args))))))
;;; ------------------------------------------------------------
;;; HTML helper macros
(defparameter *title-color* :blue-steel-light)
(defparameter *table-color* :blue-light)
(defparameter *section-title-color* :grey-dim)
(defmacro table ((stream &optional caption caption-url) &body body)
`(netscape3.0:with-table (:border nil :stream ,stream :background
,*table-color*)
(when ,caption
(row (,stream "Project" :title-background ,*title-color*
:content-alignment :center)
(netscape3.0:with-font (:size 6 :stream ,stream)
(if ,caption-url
(html3.2:note-anchor ,caption :reference ,caption-url :stream
,stream)
(princ ,caption ,stream)))))
,@body))
(defmacro row ((stream title &key title-background (content-alignment
:left)) &body content)
`(netscape3.0:with-table-row (:stream ,stream)
(netscape3.0:with-table-cell (:stream ,stream
:header-p t
:vertical-alignment :top
:horizontal-alignment :left)
(netscape3.0:with-font (:color ,*section-title-color* :stream ,stream)
(princ ,title ,stream)))
(netscape3.0:with-table-cell (:stream ,stream
:horizontal-alignment ,content-alignment
:background ,title-background)
,@content)))
;;; ------------------------------------------------------------
;;; How do we render a project description to HTML?
(defmethod render-to-html ((project project) stream)
(table (stream (project-name project) (and (slot-boundp project 'url)
(slot-value project 'url)))
(loop for section in '(description reason time-frame state steps parts
sub-projects)
when (and (slot-boundp project section)
(slot-value project section))
do (row (stream section)
(render-section-to-html project section stream)))))
(defmethod render-section-to-html ((project project) (section (eql
'sub-projects)) stream)
(loop for (nil sub-project) in (slot-value project section)
do (render-to-html sub-project stream)))
(defmethod render-section-to-html ((project project) (section (eql
'reason)) stream)
(format stream "~a" (slot-value project section)))
(defmethod render-section-to-html ((project project) (section (eql
'time-frame)) stream)
(format stream "~a" (slot-value project section)))
(defmethod render-section-to-html ((project project) (section (eql
'description)) stream)
(format stream "~a" (slot-value project section)))
(defmethod render-section-to-html ((project project) (section (eql
'state)) stream)
(format stream "~a" (slot-value project section)))
(defmethod render-section-to-html ((project project) (section (eql
'steps)) stream)
(html3.2:with-enumeration (stream :itemize)
(loop for step in (slot-value project section)
do (html3.2:enumerating-item (stream)
(princ step stream)))))
(defmethod render-section-to-html ((project project) (section (eql
'parts)) stream)
(html3.2:with-enumeration (stream :itemize)
(loop for step in (slot-value project section)
do (html3.2:enumerating-item (stream)
(princ step stream)))))
;;; ------------------------------------------------------------
;;; How does the HTML look like?
#+mcl
(defun view-html (object)
"Renders the object and shows it in the browser."
(flet ((file-to-url-string (pathname)
(concatenate 'string
"file:///"
(substitute #\/ #\:
(namestring (truename pathname))))))
(with-open-file (stream "ccl:test.html"
:direction :output
:if-exists :supersede)
(render-to-html object stream))
(open-url (file-to-url-string "ccl:test.html")
:activate-p t)))
;;; ------------------------------------------------------------
;;; Some more years...
(defproject al ("Advanced Lisp" :time-frame :long-term)
(:UI "CLIM"
:url "http://www.lavielle.com/~joswig/lisp.html#CLIM"
:reason "Don't waste your time using non-portable UI libraries"
:state :production-use
:steps ("Make it more popular"
"Finish the Specification"
"Fix the bugs"
"Free version"))
(:web "CL-HTTP"
:url "http://www.ai.mit.edu/projects/iiip/doc/cl-http/home-page.html"
:state :experimental-use
:reason "Don't waste your time using non-portable Web libraries"
:steps ("Make it more popular"
"Work on the proposed projects (see the CL-HTTP web site)"
"Make it a competetive web server on Unix"))
(:lisp "Common Lisp"
:url
"http://www.harlequin.com/education/books/HyperSpec/FrontMatter/index.html"
:state :production-use
:reason "It's there and it kind of works"
:steps ("Make it more popular"
"More standards"
"Cheaper commercial versions on Unix"
"Improve the free Common Lisps: CLisp and CMU CL"
"Motivate companies to publish their Lisp libraries"))
(:environment "CLIM Development Environment"
:url
"http://kogs-www.informatik.uni-hamburg.de/~moeller/uims-clim/clim-intro.html"
:state :development
:reason "Genera shows the potential."
:steps ("Write more code"
"Make it look really sexy"
"Integrate existing stuff"
"Port it to other platforms"))
(:internet "CL Internet Services"
:url "http://wilson.ai.mit.edu/cl-http/projects.html"
:description "Implement all kind of Internet Services in
Common Lisp"
:reason "Enables well designed and maintainable software"
:state :idea
:parts ("Basic IP protocol layer"
"Mail: POP3, SMTP, ESMTP, IMAP4"
"News"
"SNMP"
"ICMP"
"LDAP"))
(:ai "Artificial Intelligence"
:url "http://www.cs.berkeley.edu/~russell/ai.html"
:description "Use AI techniques in applications"
:reason "get better software adaptivity"
:state :research)
(:community "ALU"
:url "http://www.elwoodcorp.com/alu/table/contents.htm"
:description "Association of Lisp Users"
:steps ("Web Site"
"Provide Educational Material"
"Public Relation"
"Online Books"
"provide added values for members"
"life long email address for members: ····@lisp.org"
"programming competition: tools, applications, ..."
"telefon number: 1-800-Lisp-User"
"catalog of tools"
"comparision of Lisp tools"
"official ALU benchmark"
"official membership logo"
"T-Shirt and coffee mugs with official membership logo"
"sponsoring"
"conference"
"Repository"
"Live FAQ"
"Discussion forums"
"News/Mail Archives"
"Donations"
"Hall of Fame"))
(:platform "New Lisp Machine"
:state :idea
:reason "Most other stuff sucks!"
:steps ("Get a company/community going"
"Outline an architecture/vision"
"Setup the environment"
"Do it")))
;;; ------------------------------------------------------------
;;; Checking it
#+mcl
(view-html al)
;;; ------------------------------------------------------------
;;; End of Code
> Program large systems in Lisp.
Program large and small systems in Lisp.
--
http://www.lavielle.com/~joswig
Rainer Joswig wrote in message ...
>In article <··················@news.newsguy.com>,
>········@david-steuber.com wrote:
>
>> Linux will get better hardware support over time.
>
>A lot of stuff is getting better hardware support over time.
>
>> The more I learn about the
>> language, the more I wonder why it isn't in more general use.
>
>mostly written in CL/Scheme (more moving to CL). Some customer
>projects have been written in Lisp and we are currently
>preparing one or two new projects using Lisp, also trying to introduce
>more efficient tools.
>
>My long term strategy. O.k., let's waste a few minutes ;-) :
>
Great job but here we go again. Now there will be two niches for lisp AI
projects and writing web pages/servers. There is a thread that the
commented on the vices of basic..a very useful language in the right place
(imo). The crititcism seemed to be that one could write programs that
didn't meet some pedgogical standard. Horse feathers. There is a lisp that
will run basic into the ground, that being muLisp. Since mulisp is written
in assmebler, t'is truly quick. Not portable across platforms because a
number of functions are platform dedicated. It turns out to be very
parctical. I have written application in mulisp to control ovens temps for
1000 hrs at constant temps from reading and writing rs232 devices; EKG and
EEG input drivers for visual display and analysis. Interfaces to drive
micro-pumps, home airconditioners, secutiry devices, gps devices, drip
irragation systems, EPROM'ed mulisp in embeded projects; pharmacokinetic
calcuations, fromat and send/receive hopital patient data to foreing
systems. We even used it to link a pdp11 and ibm 4138 (that was no fun and
not to successful) ......well no use going over all 15 years. the point is i
used mulisp not really lisp. mulisp because it was a lisp dedicated to the
operating system (not to portabilty) but VERY powerful on it's platform. A
language has got to be a tool, not a set of standards.
I just upgraded my mulisp. My serial # was 942. I wonder how many VB have
been sold. Say what you want to about VB...it will let you get the job
done. I am not sure the progams will pass pedagogical scrutiny but it will
be gathering data, finding gps locations, analyzing EKG....oh well.... lisp
could be doing the same thing well maybe not all lisps. I think a language
has got to be useful before it is used. Lisp can be one of the most usefuls
tool every ran on a computer. as a matter of fact i use it, it's called
muLisp........guess what it's somewhere around $150. Last week I made that
in 3 hours of private consulting with mulisp upgrading the above oven
control program to new rs232 devices. So where does all this dribble
lead....if you want lisp to be used by JQ Public, build yourself a drip
irrigation system to water your garden and let lisp run it, you'd be shocked
how quickly your neighbors ask for one. And one more point while i'm on a
roll. I can write programs in lisp that would make the programs in basic
look like they were written at MIT.......i still get paid for them and they
run and run and run and run......goto run...... :) jump :run....catch
run:...throw run etc etc etc.
going to go to goto hell someday
rusty
In article <············@excalibur.flash.net>, "rusty craine"
<········@flash.net> wrote:
> Rainer Joswig wrote in message ...
> >In article <··················@news.newsguy.com>,
> >········@david-steuber.com wrote:
> >
> >> Linux will get better hardware support over time.
> >
> >A lot of stuff is getting better hardware support over time.
> >
> >> The more I learn about the
> >> language, the more I wonder why it isn't in more general use.
> >
> >mostly written in CL/Scheme (more moving to CL). Some customer
> >projects have been written in Lisp and we are currently
> >preparing one or two new projects using Lisp, also trying to introduce
> >more efficient tools.
> >
> >My long term strategy. O.k., let's waste a few minutes ;-) :
> >
> Great job but here we go again. Now there will be two niches for lisp AI
> projects and writing web pages/servers.
Hmm, I don't think that this should be the summary. Don't get me
wrong, this was
*******************my********************
personal strategy and wishlist. This
was what I am for my personal need and our companies needs are looking for.
It is totally subjective and probably doesn't meet your needs.
What is your long term direction?
> commented on the vices of basic..a very useful language in the right place
> (imo).
Sure.
> didn't meet some pedgogical standard. Horse feathers. There is a lisp that
> will run basic into the ground, that being muLisp.
Actually I doubt that. Several Basic dialects have a much
larger infrastructure, than for example muLisp.
> Since mulisp is written
> in assmebler, t'is truly quick.
Hmm, I think this is neither necessary or sufficient.
> operating system (not to portabilty) but VERY powerful on it's platform. A
> language has got to be a tool, not a set of standards.
Sometimes that is true sometimes not. There are larger systems
written in Lisp that do control much larger technical systems
(like factories). I guess these also have to fulfill certain
standards (like communication protocols, supporting
a certain device bus standard, ...).
> I just upgraded my mulisp. My serial # was 942. I wonder how many VB have
> been sold.
Many millions, I guess.
I don't get your point, yet. You seem to have great success with
a small dedicated Lisp system - still thousands of people are using Basic,
Assembler, C, whatever for a similar task. But, I don't doubt
that you can use Lisp and run it in small systems to
gather data and control devices (for example some robotics guys want to
do the same). I also think one should be aware of this possibility.
--
http://www.lavielle.com/~joswig
In article <············@inconnu.isu.edu>,
Craig Kelley <···@inconnu.isu.edu> wrote:
>In article <············@nnrp1.dejanews.com>, <········@acm.org> wrote:
>Heh, at least Apple is moving *toward* UNIX.
Again. I still have an Apple A/UX tee-shirt - from YEARS ago.
--
Bill Vermillion bv @ wjv.com
Followup-To:
On Thu, 05 Nov 1998 11:30:26 GMT, ········@acm.org <········@acm.org> wrote:
>
>Of course IRIX is the best Unix around
The best unix around for what? I doubt that there's any need to explain
why I take issue with this statement.
--
Ben
"You have your mind on computers, it seems."
void <·····@interport.net> wrote:
>Followup-To:
>
>On Thu, 05 Nov 1998 11:30:26 GMT, ········@acm.org <········@acm.org> wrote:
>>
>>Of course IRIX is the best Unix around
>
>The best unix around for what? I doubt that there's any need to explain
>why I take issue with this statement.
Don't bother explaining. Just suspend disbelief and consider the phrase 'united
we stand'. Nowadays, 'unix is unix' is good enough.
--
Alastair
work : ········@psoft.co.uk
home : ········@calliope.demon.co.uk