💾 Archived View for blitter.com › apl-books › APLSE › www.sigapl.org › Archives › waterloo_archive ›… captured on 2022-07-17 at 03:52:11.
-=-=-=-=-=-=-
From owner-jsoftware@lists.interlog.com Sun Aug 1 19:28:13 1999
Date: Sun, 1 Aug 1999 16:26:50 -0700
From: gordon@island.net (Ken Ian Gordon)
Subject: Re: Jforum: Viewing large arrays. User help
Brian Bambrough wrote
>A few days ago I posted some questions on the J Forum. Two of them
>were to do with wdview and "what's this?" kind of help. Nobody had
Sorry about that.
........
>
>If anyone has suggestions for improvements I would be greatly
>interested.
I used an alternative approach to supply help on buttons. As you know it is
available for menu entries, but I too could find nothing for buttons in J.
I used the F1 key instead. When pressed it ascertained which button had the
focus and displayed the appropriate short help message in the status area as
taken from a list. Using TAB to move focus and pressing F! then gave all
the help messages for the buttons. It also gave short help messages for
anything data entry fields, to explain the required entry. These were taken
from the meta data.
Not as good as right button, or the automatic display for menus, but it
appeared to suffice.
regards
kig
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Mon Aug 2 00:16:50 1999
Date: Mon, 02 Aug 1999 00:15:29 -0400
From: David Ness <DNess@home.com>
Subject: Jforum: A (surely) stupid idea ...
I find that I have (slightly) less trouble reading J than K. I should
say at the outset, that I am not bitching, here, as I find that reading
both is enormously profitable for me. But I wonder if one might help
me with the other.
For example, in the recently discussed `Combinations' problem, Roger
produced an answer that I understood after a couple of hours of
study, while over on the K-forum, Arthur's equally elegant answer took
me an extra couple of days. All of the hours and days were well spent
in both endavours, so I am not complaining about any of this. However,
in the process I did discover that, at least for me, J's ability to
`expose' the parse of the input and give me sort of a `diagrammatic'
representation of how it `sees the world' as it executes the input
is a step up on understanding why the answers that it produces are
right, and my thoughts about what it `should do' are invariably naive.
This leads me to wonder if it would be an interesting and instructive
exercise to try to produce a `K simulator' in J. I certainly understand
that producing a full K simulator would be both hard and pointless.
However, short of that, I find I have enough trouble with understanding
some of the elemental aspects of K, that `exposing' their opertaion in
J would be some contribution to my understanding of both languages.
I understand that this is probably all silly, but I have the feeling
that if the `better heads' here are able to tell me why, I'll learn
something from raising the question.
I recognize also that I could equally well raise this question in the
K forum. However, given that I have a harder time, given present
experience, understanding K, for me this would be an exercise in
trying to use the more obscure to un-obfuscate. And thus a more
difficult task...
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Mon Aug 2 15:25:30 1999
Delivered-To: fixup-forum@JSoftware.Com@fixme
Date: Mon, 02 Aug 1999 12:22:56 -0700
From: greg heil <gheil@uswest.net>
Subject: Re: Jforum: Cross Post: A (surely) stupid idea ...
References: <37A51B61.46C29812@Home.Com>
David Ness wrote:
> ... in the recently discussed `Combinations' problem, Roger
produced an answer that I understood after a couple of hours
of study, while over on the K-forum, Arthur's equally elegant
answer took me an extra couple of days.
The combinations problem can be a nice "Rosetta Stone" which
was one of the reasons i brought it up on the K-list. Other than
that i actually needed the functionality;} Rogers and Arthurs
solutions are quite parallel, too.
NB. Roger Hui:
rc=: 4 : 0 NB. All size x. combinations of i.y.
z=.1 0$k=.i.#c=.1,~(y.-x.)$0
for. i.x. do. z=.;k,.&.>(-c=.+/\.c){.&.><1+z end.
)
/Arthur Whitney:
ac:{,/x{k,''(+\#:'x)#\:,/1+x}/k:|!y-x-:1}
K derives much of its brevity in this example from the
fact that list operands are the native type, whereas in
J the under operator, &., can be combined with boxing, <,
to achieve the same end.
The k's in both implementations serve the same purpose,
to be the addend for each growing combination and are
formed analogously through index set generators.
Being as {k,''(+\#:'x)#\:,/1+x} is a monad the first x in
ac serves as a loop count, in parallel with rc's for. i.x. do.
rc's c is implicit in the ac (+\#:'x). Which is to say a scan
of the counts of the previous iteration.
Putting these together in the iterative step of the rc case:
k,.&.>(-c=.+/\.c){.&.><1+z
the addends k are laminated (,.&.>) to count (-c=.+/\.c)
copies ({.&.>)) of the incremented (<1+z) results of the
previous iteration (for. i.x. do.).
Much like in ac where k's are laminated (,'') to count (+\#:'x)
copies (#\:) of the ravelled (,/) incremented (1+x) results of
the previous iteration (x{}/).
A strictly kosher way to figure out a phrase's meaning is to
use the languages debug facility to dissect it. In K's case
this is the stop/trace/null facility. Setting it to trace
\b t / and instrumenting ac:
ac:{+,/x{\ k,''(\ +\#:'x)#\:,/1+x}/k:|!y-x-:1}
ac[3;6] / and running...
1 2 3 4 / 1st iteration counts; and results:
(,3 4
(2 4
2 3)
(1 4
1 3
1 2)
(0 4
0 3
0 2
0 1))
1 3 6 10 / 2nd iteration
(,3 4 5
(2 4 5
2 3 5
2 3 4)
(1 4 5
1 3 5
1 3 4
1 2 5
1 2 4
1 2 3)
(0 4 5
0 3 5
0 3 4
0 2 5
0 2 4
0 2 3
0 1 5
0 1 4
0 1 3
0 1 2))
/ and final result (transposed for brevity)
(3 2 2 2 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0
4 4 3 3 4 3 3 2 2 2 4 3 3 2 2 2 1 1 1 1
5 5 5 4 5 5 4 5 4 3 5 5 4 5 4 3 5 4 3 2)
> This leads me to wonder if it would be an interesting and
instructive exercise to try to produce a `K simulator'...
Stevan Apter has put out an interpreter for "pure K" a couple
times (eg DejaNews for comp.lang.apl). But that is still a
long way from capturing all the syntactic sweeteners.
greg heil
mailto:gheil@acm.org
http://www.scn.org/tl/anvil
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Wed Aug 4 01:51:59 1999
From: k-list@iname.com
Date: Wed, 4 Aug 1999 01:47:24 -0400 (EDT)
Content-Type: Text/Plain
Subject: Jforum: implementation of "record lock"
"bill lam"<k-list@iname.com>
I'm trying to implement an address book application using J4.03,
it uses ddsel(select *) to fetch a record for view/change,
and then uses ddsql(update) for updating. It works fine, until
concurrent access occurs. The obvious solution is of course using
"record lock", but how? I can think of 2 possible ways,
1) use ODBC "cursor library"
2) use a centralised file for holding record lock information.
but I haven't tried these yet, anyone has experience in this
subject matter or other suggestions?
---------------------------------------------------
Get free personalized email at http://www.iname.com
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Wed Aug 4 08:40:04 1999
From: gosi@centrum.is
Subject: Re: Jforum: implementation of record lock
Date: Wed, 4 Aug 99 12:38:57 +0000
k-list@iname.com writes:
> "bill lam"
> I'm trying to implement an address book application using J4.03,
> it uses ddsel(select *) to fetch a record for view/change,
> and then uses ddsql(update) for updating. It works fine, until
> concurrent access occurs. The obvious solution is of course using
> "record lock", but how? I can think of 2 possible ways,
> 1) use ODBC "cursor library"
> 2) use a centralised file for holding record lock information.
It all depends on the database you are using and what
options that database allows you to do.
if your application is the only application using this
database then of course you can use your own
methods. Otherwise it is up to the database.
/Gosi
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Wed Aug 4 17:17:56 1999
From: "Nichols, Peter" <pnichols@sprinc.com>
Subject: Jforum: TreeView OCX
Date: Wed, 4 Aug 1999 15:46:12 -0500
Could someone give me an example of using the TreeView OCX? And also what
other tree controls are available and any recommendations would be great.
Thanks,
Pete Nichols
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Aug 5 06:23:57 1999
From: <paul_gauthier@dynasys.tm.fr>
Subject: Jforum: A chanllenge...
Date: Thu, 5 Aug 1999 12:21:25 +0200
I have a future doctor in Operational Research who beleives that no one
in the world can achieve better results than the CPLEX corporation
solver on integer's linear programming.
Anyone interested to give hints on a better solution please contact me.
Regards/Paul
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Aug 5 11:37:39 1999
From: "Oleg Kobchenko" <gccinc@usa.net>
References: <01074BD4EEC4D1118E3200805F6542A7177FF9@SPRDALLAS2>
Subject: Re: Jforum: TreeView OCX
Date: Thu, 5 Aug 1999 18:38:39 +0300
charset="koi8-r"
You can check http://members.xoom.com/olegyk/jpage.html
Also J distribution should have such sample.
TreeView should saticefy most needs. A single
example of a different and original type of
tree is Xerox Hyperbolic Tree control.
----- Original Message -----
From: Nichols, Peter <pnichols@sprinc.com>
Sent: Wednesday, August 04, 1999 23:46
Subject: Jforum: TreeView OCX
> Could someone give me an example of using the TreeView OCX? And also what
> other tree controls are available and any recommendations would be great.
>
> Thanks,
> Pete Nichols
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Aug 5 14:30:50 1999
From: btanksley@hifn.com
Subject: RE: Jforum: A chanllenge...
Date: Thu, 5 Aug 1999 11:29:15 -0700
charset="iso-8859-1"
Could you clarify? I've never heard of that solver, but my professor
specializes in integer programming.
-Billy
>-----Original Message-----
>From: paul_gauthier@dynasys.tm.fr [mailto:paul_gauthier@dynasys.tm.fr]
>Sent: Thursday, August 05, 1999 3:21 AM
>To: forum@jsoftware.com
>Subject: Jforum: A chanllenge...
>
>
>I have a future doctor in Operational Research who beleives that no one
>in the world can achieve better results than the CPLEX corporation
>solver on integer's linear programming.
>
>Anyone interested to give hints on a better solution please contact me.
>
>Regards/Paul
>
>---------------------------------------------------------------
>-----------------
>J Forum: for information about this list, see
>http://www.jsoftware.com/forum.htm
>
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Aug 7 01:25:16 1999
Date: Fri, 06 Aug 1999 23:40:48 -0400
From: Brian Bambrough <b.bambrough@worldnet.att.net>
Subject: Re: Jforum: Viewing large arrays. User help
References: <199908012326.QAA25635@mimas.island.net>
Ken Ian Gordon wrote:
> I used an alternative approach to supply help on buttons. As you
> know it is available for menu entries, but I too could find nothing
> for buttons in J. I used the F1 key instead. When pressed it
> ascertained which button had the focus and displayed the appropriate
> short help message in the status area as taken from a list. Using
> TAB to move focus and pressing F1 then gave all the help messages
> for the buttons. It also gave short help messages for anything data
> entry fields, to explain the required entry. These were taken from
> the meta data.
> Not as good as right button, or the automatic display for menus, but
> it appeared to suffice.
The big advantage of your approach is that there are no extra controls
in the form. In my approach I have to add a "Button help" check box.
As I understand it, in your method the only way to set the focus on a
button is to tab to it. You can't set the focus with a mouse click as
this would invoke the action of the button. Is this right?
Martin Neitzel wrote:
> re your enquiry about context sensitive help: In the J Workbench, I
> use the following approach to jump around between verb definitions:
> If the "goto verb" function is triggered (by using Ctrl-G or
> clicking a button), I make a check whether there is some text
> selection in one of several text fields. If so, I extend the
> partial selection to the whole word near to the left selection
> border and look that up in my list of identifiers. If I can find
> it, I jump to it. Otherwise (no match or no selection) I pop up a
> menu.
> This works quite well, in particular because ANY text output
> benefits from this jump feature (unlike html-text which needs all
> links to be prepared). For example, users are able to jump quickly
> to a verb listed in a boxed representation.
> While this is a very specific context for "jumps", and your question
> for Context Sensitive Help solutions appears to be much more
> general, I thought I might drop short note.
I must confess that I don't grasp what you wrote, although it sounds
very interesting. I don't know what a "J Workbench" is. Is it a tool
for developing in J? If so, I would like to know more about it. I will
be at APL99 next week. Perhaps we could talk about it then.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Aug 7 13:43:43 1999
Date: Sat, 7 Aug 1999 10:38:40 -0700
From: gordon@island.net (Ken Ian Gordon)
Subject: Re: Jforum: Viewing large arrays. User help
Brian Bambrough wrote
>Ken Ian Gordon wrote:
>
>> I used an alternative approach to supply help on buttons. As you
>> know it is available for menu entries, but I too could find nothing
>> for buttons in J. I used the F1 key instead. When pressed it
<snip>
>
>The big advantage of your approach is that there are no extra controls
>in the form. In my approach I have to add a "Button help" check box.
>
>As I understand it, in your method the only way to set the focus on a
>button is to tab to it. You can't set the focus with a mouse click as
>this would invoke the action of the button. Is this right?
Unfortunatley yes.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Aug 7 22:29:45 1999
From: Ajith Prasad <burap@nus.edu.sg>
Subject: Jforum: KEYED FILES & DICTIONARIES
Date: Sun, 8 Aug 1999 10:24:17 +0800
A recent discussion forum resulted in a script ('jzdict')for handling of
"Dictionaries" being included in the "system\classes\dict" directory. I have
just noticed that J also includes a 'kfiles' script in the
"system\packages\files" directory which allows data to be accessed using
keywords. The functionality seems similar to that in the dictionary script.
Has anyone tried both scripts and prefers one to the other?
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Aug 8 04:37:39 1999
From: "Stefano Lanzavecchia" <stf@apl.it>
References: <A12A4ACD69B4D011BB960020AFFBF108029E931D@exs02.ex.nus.edu.sg>
Subject: Re: Jforum: KEYED FILES & DICTIONARIES
Date: Sun, 8 Aug 1999 10:27:32 +0200
Organization: APL Italiana
charset="iso-8859-1"
----- Original Message -----
From: Ajith Prasad <burap@nus.edu.sg>
Sent: Sunday, August 08, 1999 4:24 AM
Subject: Jforum: KEYED FILES & DICTIONARIES
: A recent discussion forum resulted in a script ('jzdict')for handling of
: "Dictionaries" being included in the "system\classes\dict" directory. I
have
: just noticed that J also includes a 'kfiles' script in the
: "system\packages\files" directory which allows data to be accessed using
In two words: "kfiles" is a component file system (like APL's traditional)
where the components are accessed not by number but by name. "jzdict" keeps
the dictionary in memory, dictionary which can be exported to file and
re-imported from file if the user wants it.
--
WildHeart'99
Homepage: http://www.insight.dk/stf/
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Aug 8 12:42:24 1999
From: "Anne Faust" <amfaust@mindspring.com>
Subject: Jforum: Press Release - Iverson Software Inc.
Date: Sun, 8 Aug 1999 12:45:30 -0400
charset="iso-8859-1"
Eric Iverson has been appointed Chief Technical Officer with Adaytum
Software. See http://www.adaytum.com for information about Adaytum.
In addition to being attracted to Adaytum's business, rapid growth, and
dynamic company culture, Eric says that a primary reason for joining is
Adaytum's commitment to J as a key technology in building the future for
Adaytum and its clients.
Adaytum's teams of J developers working on building new products is a
significant step forward for J and will help ISI build and maintain J
momentum.
In addition, the ISI development team are planning the following upcoming
releases. Later this year there will be a release that supports sparse
arrays. Roger Hui will present a paper on this at the upcoming APL99
conference, see http://www.lingo.com/apl99. Also, early next year there will
be a release that includes a Session Manager and Window driver for J on
Unix platforms with an integrated set of libraries and utilities allowing
many complete applications to move seamlessly between Windows and Unix.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Aug 8 14:38:45 1999
From: "Gordon Sutcliffe" <gsutcliffe@talk21.com>
Subject: Jforum: Use of new Plot option types: Linefit and Step
Date: Thu, 5 Aug 1999 15:18:29 +0100
boundary="----=_NextPart_000_0018_01BEDF55.C5D2E6E0"
This is a multi-part message in MIME format.
------=_NextPart_000_0018_01BEDF55.C5D2E6E0
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
The following J 4.03 script (Windows 95), displays exception messages on =
the linefit and step options of the plot command:
=20
dsp =3D: 1!:2&2 NB. function to copy left argument to display
errmsg =3D: 13!:12
=20
require 'plot trig'
tp =3D: 3 : 0 NB. Test Plot for all plot types listed in User Manual
t =3D. 'area bar density dot errorbar hilo line linefit fbar sbar '
t =3D. <;._2 t, 'pie point radar stick step surface wire '
=20
for_i. t do.
dsp i=3D.>i NB. dislay plot option type
try. ('type ', i, ';') plot 3 6$i.6
catch. dsp (i, ': fails with error message:'),: errmsg ''
end.
6!:3 y.
end.
0 0