Vector Volume 23 N°3

29 views
Skip to first unread message

Stephen Taylor <editor@vector.org.uk>

unread,
Apr 24, 2008, 2:30:40 PM4/24/08
to
Two more articles from my in-tray have made it online.

Keith Smillie fancied a Differential Analyzer build from Meccano
parts, but found it easier just to model it.

Cliff Reiter takes his GPS thingy hiking. When he gets back, his
program lets him know where he's been.

Riveting narrative, arresting pictures and real, executable source
code – all at

http://www.vector.org.uk/archive/v233/smillie.htm
http://www.vector.org.uk/archive/v233/reiter.htm

Stephen Taylor
edi...@vector.org.uk

AAsk

unread,
Apr 25, 2008, 1:44:07 AM4/25/08
to
I had a quick look at the Reiter article: what language is the code in?

Gosi

unread,
Apr 25, 2008, 3:48:55 AM4/25/08
to
On Apr 25, 5:44 am, AAsk <AA2e...@lycos.co.uk> wrote:
> I had a quick look at the Reiter article: what language is the code in?

J

from:

http://jsoftware.com/


phil chastney

unread,
Apr 25, 2008, 4:06:30 AM4/25/08
to
Stephen Taylor <edi...@vector.org.uk> wrote:
>
> Keith Smillie fancied a Differential Analyzer build from Meccano
> parts, but found it easier just to model it.

golly -- I never built it it, mind, but I've still got the original
plans for that

reading through said plans, it transpired that the thing was accurate to
about 1 or 2 sig figs, the problem being the differential unit itself
which consisted of a rubber shod wheel being rotated by a flat
(cardboard?) disc operating in a plane tangential to the rim of the wheel

basically, the wheel slipped if the flat disc was rotating too fast

I used to use a Brunsviga during the early days of an abortive actuarial
career, but in the end I gave up on mechanical means of computation

Heathkit used to do a kit for an analogue computer -- something else I
never built

nowadays, of course, it's all FFT and DSP . . . /phil

AAsk

unread,
Apr 25, 2008, 7:23:02 AM4/25/08
to
So much for readability without APL symbols then!!

jk

unread,
Apr 25, 2008, 7:50:07 AM4/25/08
to

"AAsk" <AA2...@lycos.co.uk> wrote in message
news:9f26a555-8ee9-4991...@t54g2000hsg.googlegroups.com...

> So much for readability without APL symbols then!!

what's readable on 20.000 lines of COBOL or Fortran?


Ric

unread,
Apr 25, 2008, 8:42:40 AM4/25/08
to
On Apr 25, 11:23 pm, AAsk <AA2e...@lycos.co.uk> wrote:
> So much for readability without APL symbols then!!

A little over a year ago that may have been my response too. But now I
would say that, just like with APL (and any other language that uses a
previously unknown set of symbols) there is some initial work involved
in learning the symbols of the language.

Right now you have already learnt the APL symbol set and for that
reason you find it to be readable. If you'd put the same effort into
learning the J symbol set you'd probably find it just as readable as
APL. One thing I did find easier about learning the J symbol set was
that I already knew which how to type the symbols.

Curtis A. Jones

unread,
Apr 25, 2008, 2:24:20 PM4/25/08
to
> > Keith Smillie fancied a Differential Analyzer build from Meccano
> > parts, but found it easier just to model it.
...
Here's a Meccano differential analyzer.
http://meccano.us/differential_analyzers/robinson_da/index.html

Tim Robinson, who built it, also built a couple of Babbage difference
engines. e.g.
http://www.meccano.us/difference_engines/rde_2/index.html
Tim is in thick of preparations at the Computer History Museum in
Mountain View, California, USA to show a 5-ton Difference Engine no. 2
built by the Science Museum of London:
http://www.computerhistory.org/babbage/

http://www.computerhistory.org/virtualvisiblestorage/popup_image.php?base_name=102637025
It's more impressive than one might expect from the picture.

If you're around London don't despair if you want to see it. The
Babbage engine at the Computer History Museum is the second one built
by the Science Museum and is brand new.
Curtis

P.S.
Tim used Mathematica, with its extended-precision arithmetic, to
generate the starting values to put into the engine to demonstrate use
of the difference engine to make a log table which can be compared to
one published by Babbage himself.

Gosi

unread,
Apr 25, 2008, 4:18:05 PM4/25/08
to

Just like with APL chars then most people only master a handful of the
operations and use them over and over.
Some gradually learn a few more as time goesby.
Unfortunately we do lose most potential users before they learn enough
to get by.

What I really do like about J is the extensive set of labs and demos
that you can take and change slightly to fit your needs.
I was forunate to be forced into J because I could not use the APL
chars.

I have very recently been reintroduced with the APL chars through
Dialog 12 Unicode and I do like it.

Both APL with chars in Unicode and J have a very great opportunity to
grow now.
I think both are as robust and good now as the sales people said they
were decades ago and APL was not ready then.
It was also then a beautiful tool in the hands of experts but a
disaster in the hands of novices that managed to bring the biggest
computers to its knees doing strange worthless stuff.
It should have come with a warning then.
It could and often did brake down with silly warnings and telling
anyone that it was APL even if it was silly programmers fault.

phil chastney

unread,
Apr 25, 2008, 7:52:13 PM4/25/08
to

my goodness, that stuff is gorgeous -- thanks ever so much for these links

to the best of my knowledge, the red and green Meccano is no longer
available -- the company is now French-owned, I believe, and the colours
tend to be more (what..?) fashionable? purple and silver, for instance

IIRC, when the Science Museum built their first difference engine, they
found that Babbage's problem was not only that the required accuracy was
not available at that time, but the materials would not have been up to
the job either -- I was quite surprised that something as simple as
brass could vary so much -- and the weight! 5 tons is an awful lot of
metal to move without a proper supply of power

/phil

phil chastney

unread,
Apr 25, 2008, 7:57:31 PM4/25/08
to
Gosi wrote:
>
> It was also then a beautiful tool in the hands of experts but a
> disaster in the hands of novices

much the same is true of F1 cars and Concorde and neurosurgery

I never understood the argument that technology should be restrained
until your average Joe Soap could comprehend it

/phil

Bob Cain

unread,
Apr 26, 2008, 4:40:56 AM4/26/08
to
AAsk wrote:
> So much for readability without APL symbols then!!

Has anyone has tried to devise a symbol set especially for J so that its
programs don't look so much like they were written by a cat on a hot keyboard?


Bob
--

"Things should be described as simply as possible, but no simpler."

A. Einstein

Gosi

unread,
Apr 26, 2008, 5:16:50 AM4/26/08
to

Beauty is in the eye of the beholder

Curtis A. Jones

unread,
Apr 26, 2008, 11:15:26 AM4/26/08
to
On Apr 25, 4:52 pm, phil chastney
<phil.hates.s...@amadeus.munged.eclipse.co.uk> wrote:
...

> my goodness, that stuff is gorgeous -- thanks ever so much for these links
>
> to the best of my knowledge, the red and green Meccano is no longer
> available -- the company is now French-owned, I believe, and the colours
> tend to be more (what..?) fashionable? purple and silver, for instance
>
> IIRC, when the Science Museum built their first difference engine, they
> found that Babbage's problem was not only that the required accuracy was
> not available at that time, but the materials would not have been up to
> the job either ...
>
> /phil

Phil,
Tim Robinson would say that if a working difference engine can be
built from Meccano, tolerances must not have been real critical to the
design. One problem when the first difference engine was being built
in Babbage's time (ca. 1830) was his engineer/machinist Joseph
Clement* was a perfectionist who may have taken more time than needed
achieving higher tolerances than were needed in many parts of the
machinery. The Science Museum of London, when building Difference
Engine #2 used, as much as it could determine, materials that would
have been used in Babbage's time. It is true that getting everything
to work smoothly is a major hassle! Curtis

* Interchangeable parts were on the near horizon when Babbage and
Clement worked on the difference engine. One of Clement's employees
was ????? Whitworth whose name persists in screw thread nomenclature.

Gosi

unread,
Apr 26, 2008, 11:43:04 AM4/26/08
to
On Apr 25, 11:52 pm, phil chastney

<phil.hates.s...@amadeus.munged.eclipse.co.uk> wrote:
> Curtis A. Jones wrote:
> >>> Keith Smillie fancied a Differential Analyzer build from Meccano
> >>> parts, but found it easier just to model it.
> > ...
> > Here's a Meccano differential analyzer.
> >http://meccano.us/differential_analyzers/robinson_da/index.html
>
> > Tim Robinson, who built it, also built a couple of Babbage difference
> > engines.  e.g.
> >http://www.meccano.us/difference_engines/rde_2/index.html
> > Tim is in thick of preparations at the Computer History Museum in
> > Mountain View, California, USA to show a 5-ton Difference Engine no. 2
> > built by the Science Museum of London:
> >http://www.computerhistory.org/babbage/
>
> >http://www.computerhistory.org/virtualvisiblestorage/popup_image.php?...

> > It's more impressive than one might expect from the picture.
>
> > If you're around London don't despair if you want to see it.  The
> > Babbage engine at the Computer History Museum is the second one built
> > by the Science Museum and is brand new.
> > Curtis
>
> > P.S.
> > Tim used Mathematica, with its extended-precision arithmetic, to
> > generate the starting values to put into the engine to demonstrate use
> > of the difference engine to make a log table which can be compared to
> > one published by Babbage himself.
>
> my goodness, that stuff is gorgeous -- thanks ever so much for these links
>
> to the best of my knowledge, the red and green Meccano is no longer
> available -- the company is now French-owned, I believe, and the colours
> tend to be more (what..?) fashionable?  purple and silver, for instance

http://www.meccano.com/about/index.php
http://en.wikipedia.org/wiki/Meccano

Bob Cain

unread,
Apr 26, 2008, 4:30:00 PM4/26/08
to

Indeed.

Gosi

unread,
Apr 26, 2008, 7:20:15 PM4/26/08
to

I have had to read a lot of code in various languages and the beauty
of the code can vary.
After a while working with a big application written by several people
you begin to recognize who wrote what just by noticing the coding
style.

It does not matter if it is C, VB, APL or J
Badly written code and especially buggy and lapped up code more or
less always begins to look ugly.
I am seldom impressed by long onliners because they are often cramped
and difficult to understand.
On the other hand some oneliners are pure gems.

Stephen Taylor <editor@vector.org.uk>

unread,
May 22, 2008, 3:27:13 PM5/22/08
to
On Apr 24, 8:30 pm, "Stephen Taylor <edi...@vector.org.uk>"

Simon Marsden's tutorial on user-defined classes in APLX 4 is now
online at

http://www.vector.org.uk/archive/v233/marsden.htm

Stephen Taylor
edi...@vector.org.uk

Stephen Taylor <editor@vector.org.uk>

unread,
May 23, 2008, 4:45:43 AM5/23/08
to
Neville Holmes' essay "Tacit J and I" is now online at

http://www.vector.org.uk/archive/v233/tji.htm

Stephen Taylor
edi...@vector.org.uk

Stephen Taylor <editor@vector.org.uk>

unread,
May 27, 2008, 7:29:46 AM5/27/08
to
On Apr 24, 7:30 pm, "Stephen Taylor <edi...@vector.org.uk>"

<StephenTaylorF...@googlemail.com> wrote:
> Two more articles from my in-tray have made it online.
>
> Keith Smillie fancied a Differential Analyzer build from Meccano
> parts, but found it easier just to model it.
>
> Cliff Reiter takes his GPS thingy hiking. When he gets back, his
> program lets him know where he's been.
>
> Riveting narrative, arresting pictures and real, executable source
> code – all at
>

Beau Webber’s review of APLX v4 is now online at

http://www.vector.org.uk/archive/v233/smillie.htm

I’m particularly pleased to be publishing this review by a non-
professional programmer – a ‘domain expert’ who in his own field/s
relies on APL to accomplish what he could not otherwise get done. An
encouraging testament. (And, yes, APLX v4 looks pretty good!)

Stephen Taylor
edi...@vector.org.uk

Kerry Liles

unread,
May 27, 2008, 9:05:05 AM5/27/08
to
"Stephen Taylor <edi...@vector.org.uk>" <StephenT...@googlemail.com>
wrote in message
news:6f5bffe6-adac-4baf...@z72g2000hsb.googlegroups.com...

Beau Webber’s review of APLX v4 is now online at

http://www.vector.org.uk/archive/v233/smillie.htm

I’m particularly pleased to be publishing this review by a non-
professional programmer – a ‘domain expert’ who in his own field/s
relies on APL to accomplish what he could not otherwise get done. An
encouraging testament. (And, yes, APLX v4 looks pretty good!)

Stephen Taylor
edi...@vector.org.uk

===================================================
I think this is the correct link:

http://www.vector.org.uk/archive/v233/webber.htm


Stephen Taylor <editor@vector.org.uk>

unread,
May 27, 2008, 2:44:20 PM5/27/08
to
On Apr 24, 7:30 pm, "Stephen Taylor <edi...@vector.org.uk>"
<StephenTaylorF...@googlemail.com> wrote:
> Two more articles from my in-tray have made it online.
>
> Keith Smillie fancied a Differential Analyzer build from Meccano
> parts, but found it easier just to model it.
>
> Cliff Reiter takes his GPS thingy hiking. When he gets back, his
> program lets him know where he's been.
>
> Riveting narrative, arresting pictures and real, executable source
> code – all at
>

Adrian Smith’s report of the APL moot in San Quirico d’Orcia last June
is now online at

http://www.vector.org.uk/archive/v233/sanquirico.htm

It includes John Scholes’ notes from his talk on
hypersuperduperoperators.

Stephen Taylor
edi...@vector.org.uk
http://www.vector.org.uk
http://aplteam2.com/aplwiki/

Stephen Taylor <editor@vector.org.uk>

unread,
May 27, 2008, 4:04:11 PM5/27/08
to
On Apr 24, 7:30 pm, "Stephen Taylor <edi...@vector.org.uk>"
<StephenTaylorF...@googlemail.com> wrote:
> Two more articles from my in-tray have made it online.
>
> Keith Smillie fancied a Differential Analyzer build from Meccano
> parts, but found it easier just to model it.
>
> Cliff Reiter takes his GPS thingy hiking. When he gets back, his
> program lets him know where he's been.
>
> Riveting narrative, arresting pictures and real, executable source
> code – all at
>

The Minnowbrook conferences revived… Adrian Smith’s report of
Minnowbrook 2007 is now online at

http://www.vector.org.uk/archive/v233/minnowbrook.htm

WildHeart

unread,
May 28, 2008, 3:38:40 AM5/28/08
to
> (And, yes, APLX v4 looks pretty good!)

I agree! APLX v4 is really cool! There's only one thing that scares
me: all their OO mechanism is based on reference counting and garbage
collectors based on ref counting cannot reclaim circular structures.
Even the simple (pseudocode follows):

father<-[]NEW object
son<-[]NEW object
father.child<-son
son.parent<-father
[]ex 'father'
[]ex 'son'

will leak because the two objects (though unreacheable from the
outside world) will keep each other alive. I find myself building
structures like those all the time...

As far as I know, Dyalog had to enhance their garbage collector with a
special pass to make sure that unreacheable instances, even though
they're alive from a purely ref count point of view, are disposed of.
-- Stef

micr...@microapl.demon.co.uk

unread,
May 28, 2008, 4:45:24 AM5/28/08
to

It's OK, Stef. APLX uses reference counts augmented by a special mark-
and-sweep pass to detect and dispose of dead circular references (this
is similar to what Python does, I believe). As it is potentially a
compute-intensive operation, we have various checks in place to ensure
that we do this mask-and-sweep pass only if it is necessary.

Richard Nabavi
MicroAPL Ltd

Stephen Taylor <editor@vector.org.uk>

unread,
May 28, 2008, 1:30:47 PM5/28/08
to
On Apr 24, 7:30 pm, "Stephen Taylor <edi...@vector.org.uk>"
<StephenTaylorF...@googlemail.com> wrote:

Adrian Smith’s article "Using the .Net DataSet Class with Dyalog 12"
has just been posted at

http://www.vector.org.uk/archive/v233/smithdat.htm

It describes how to exploit the .Net DataSet class to get a usable
DBMS without having to install SQL Server or its like. As a bonus, the
permanent store can be an XML document, editable in a text editor: all
you would need for many small applications.

WildHeart

unread,
May 29, 2008, 3:48:12 AM5/29/08
to
> > I agree! APLX v4 is really cool! There's only one thing that scares
> > me: all their OO mechanism is based on reference counting and garbage
> > collectors based on ref counting cannot reclaim circular structures.

> It's OK, Stef.  APLX uses reference counts augmented by a special mark-


> and-sweep pass to detect and dispose of dead circular references (this
> is similar to what Python does, I believe).  As it is potentially a
> compute-intensive operation, we have various checks in place to ensure
> that we do this mask-and-sweep pass only if it is necessary.

This is very cool! Thank you for making this clear.
You have a wonderful little/big product there ^__^

-- Stef

Stephen Taylor <editor@vector.org.uk>

unread,
May 30, 2008, 12:19:10 PM5/30/08
to
A brief note in the “In Session” series, on representing tables as
namespaces of vectors, has just been posted at

http://www.vector.org.uk/archive/v233/insession.htm

It contains some examples of some fairly ‘classic’ and non-trvial
relational table operations described in APL, perhaps more clearly
than in SQL.

Stephen Taylor

edi...@vector.org.uk
http://www.vector.org.uk
http://aplwiki.aplteam.com

J.B.W....@kent.ac.uk

unread,
Jun 3, 2008, 9:00:02 AM6/3/08
to
On May 27, 2:05 pm, "Kerry Liles" <kerryli...@rogers.com> wrote:
> "Stephen Taylor <edi...@vector.org.uk>" <StephenTaylorF...@googlemail.com>
> wrote in messagenews:6f5bffe6-adac-4baf...@z72g2000hsb.googlegroups.com...

>
> Beau Webber’s review of APLX v4 is now online at
>
> (see below)

>
> I’m particularly pleased to be publishing this review by a non-
> professional programmer – a ‘domain expert’ who in his own field/s
> relies on APL to accomplish what he could not otherwise get done. An
> encouraging testament. (And, yes, APLX v4 looks pretty good!)
>
> Stephen Taylor
> edi...@vector.org.uk
>
> ===================================================
> I think this is the correct link:
>
> http://www.vector.org.uk/archive/v233/webber.htm

Hi, if you want to see an example of a program coded in an earlier
version of APLX,
please see :
http://www.lab-tools.com/instrumentation/StereoPanorama/

This is a simple viewer / immediate mode editor for viewing colour
stereo panoramas,
that I knocked up at the time of the 2004 Nasa Mars rover landings.
Should be good for the current Phoenix lander when the colour stereo
images come out.
You can zoom into and pan around inside full colour stereo pictures.

APLX now has greatly enhanced graphical routines that call on
ImageMagicK,
so the program is in urgent need of updating - when I have the
time ....
cheers,
Beau

Reply all
Reply to author
Forward
0 new messages