💾 Archived View for blitter.com › apl-books › APLSE › www.sigapl.org › Archives › waterloo_archive ›… captured on 2024-08-18 at 19:33:47.
⬅️ Previous capture (2022-07-17)
-=-=-=-=-=-=-
From owner-jsoftware@lists.interlog.com Sun Jan 31 18:45:36 1999
Organization: S&PE Telematika
From: "ASB" <asb@lnk.tm.odessa.ua>
Subject: Jforum: matrix rotating
Date: Mon, 1 Feb 1999 01:32:35 +0200
In APL I can rotate rows of matrix independently.
Can I do this in J in some simple way?
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Jan 31 18:57:58 1999
Delivered-To: fixup-forum@jsoftware.com@fixme
Date: Sun, 31 Jan 1999 15:49:05 -0800
From: greg heil <gheil@uswest.net>
Subject: Re: Jforum: matrix rotating
References: <199901312335.BAA12000@Tele.TM.Odessa.UA>
ASB wrote:
> In APL I can rotate rows of matrix independently.
Can I do this in J in some simple way?
Try variants of
1 2 3 |."0 1 i.3 5
1 2 3 4 0
7 8 9 5 6
13 14 10 11 12
Key is " (rank) operator.
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 Sun Jan 31 19:53:56 1999
From: "Jay Daulton" <jovenova@earthlink.net>
Subject: Jforum: Re: wdview
Date: Sun, 31 Jan 1999 16:53:49 -0800
charset="iso-8859-1"
Is there an alternative script to the supplied windows text viewer -
wdview - that
can handle more than 20K bytes? I have an application that would require
upwards of 3 megs.
I am using J3.05b.
Jay Daulton
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Jan 31 23:51:16 1999
From: "John D. Baker" <bakerjd@kos.net>
Subject: Re: Jforum: Re: wdview
Date: Sun, 31 Jan 1999 22:34:11 -0500
charset="iso-8859-1"
The text viewer uses a standard windows control and the
amount of text it can handle is os dependent. For win 3.1x, 95/98
20k sounds about right. Under NT you can safely go up to 2megs.
For 3megs you need to get ahold of a good windows text editor that
can be be run by winexec commands or embedded as an
ole/activeX control on a J form.
-----Original Message-----
From: Jay Daulton <jovenova@earthlink.net>
Date: Sunday, January 31, 1999 7:48 PM
Subject: Jforum: Re: wdview
>Is there an alternative script to the supplied windows text viewer -
> wdview - that
>can handle more than 20K bytes? I have an application that would require
>upwards of 3 megs.
>I am using J3.05b.
>
>
>Jay Daulton
>
>
>---------------------------------------------------------------------------
-----
>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 Mon Feb 1 11:56:14 1999
References: <016001be4ba3$254cdfa0$360b14d1@f3nbp>
Organization: Voxel
From: "Andrew Nikitin" <nsg@voxel.kharkov.ua>
Date: Mon, 1 Feb 1999 13:21:19 +0200 (EET)
Subject: Jforum: @: and [:
10 ([: >: <.@^.)1208021x
7
10 (>: @: <.@^.)1208021x
|domain error
| 10 (>:@:<.@^.)1208021
After some meditation i have decided that such behaviour is completely
correct, but it still looks some puzzling.
BTW, this example shows that composition in J isn't associative as it is in
mathematic, so, be careful.
nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Mon Feb 1 12:15:16 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: @: and [:
Date: Mon, 1 Feb 1999 11:58:47 -0500
Why is this behavior correct? Since
10 ^. 1208021x
|domain error
I would have thought that 10 ([: >: <.@^.)1208021x
would fail too. Why doesn't it? It looks like the
1208021x is being converted to 1208021, but I don't
see why.
Henry Rich
-----Original Message-----
From: Andrew Nikitin <nsg@voxel.kharkov.ua>
Date: Monday, February 01, 1999 11:39
Subject: Jforum: @: and [:
>
> 10 ([: >: <.@^.)1208021x
>7
> 10 (>: @: <.@^.)1208021x
>|domain error
>| 10 (>:@:<.@^.)1208021
>
>After some meditation i have decided that such behaviour is completely
>correct, but it still looks some puzzling.
>
>BTW, this example shows that composition in J isn't associative as it is in
>mathematic, so, be careful.
>
>
>nsg
>
>
>
>---------------------------------------------------------------------------
-----
>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 Mon Feb 1 12:53:14 1999
Date: Mon, 1 Feb 1999 19:38:45 +0200 (EET)
From: Oleg Kobchenko <gccinc@usa.net>
Subject: Re: Jforum: @: and [:
In-Reply-To: <001401be4e04$250876e0$f7a956d1@seymourg>
Also ineteresting is
10 ^.&<. 1208021
6.08207
10 ^.&<. 1208021x
|domain error
| 10 ^.&<.1208021
On Mon, 1 Feb 1999, Seymour Glass wrote:
> Why is this behavior correct? Since
>
> 10 ^. 1208021x
> |domain error
>
> I would have thought that 10 ([: >: <.@^.)1208021x
> would fail too. Why doesn't it? It looks like the
> 1208021x is being converted to 1208021, but I don't
> see why.
>
> Henry Rich
>
> -----Original Message-----
> From: Andrew Nikitin <nsg@voxel.kharkov.ua>
> To: forum@JSoftware.Com <forum@JSoftware.Com>
> Date: Monday, February 01, 1999 11:39
> Subject: Jforum: @: and [:
>
>
> >
> > 10 ([: >: <.@^.)1208021x
> >7
> > 10 (>: @: <.@^.)1208021x
> >|domain error
> >| 10 (>:@:<.@^.)1208021
> >
> >After some meditation i have decided that such behaviour is completely
> >correct, but it still looks some puzzling.
> >
> >BTW, this example shows that composition in J isn't associative as it is in
> >mathematic, so, be careful.
> >
> >
> >nsg
> >
> >
> >
> >---------------------------------------------------------------------------
> -----
> >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
>
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Mon Feb 1 13:32:48 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: @: and [:
Date: Mon, 1 Feb 1999 13:19:24 -0500
I see now. The key is in Dictionary II.G para. 1, which says that '<.@f and
>.@f produce entended integer results when applied to extended integer
arguments'.
So
10 ^. 1000001x
|domain error
| 10 ^.1000001
10 ^. 1000000x
6
10 <.@^. 1000001x
6
but:
10 <.@:^. 1000001x
|domain error
| 10 <.@:^.1000001
Here is another interesting tidbit:
10 (^. 1001"_) 1001x
3.00043
I expected this to fail, because 10 would be converted to extended by the
larger verb and then applied to ^. . But apparently this didn't happen.
Dictionary II.G says 'dyadic functions...convert their arguments to the same
type'. Apparently the conversion is performed only by the 'primitive
functions'
referred to earlier in the paragraph.
Henry Rich
-----Original Message-----
From: Oleg Kobchenko <gccinc@usa.net>
Date: Monday, February 01, 1999 12:39
Subject: Re: Jforum: @: and [:
>Also ineteresting is
>
> 10 ^.&<. 1208021
>6.08207
> 10 ^.&<. 1208021x
>|domain error
>| 10 ^.&<.1208021
>
>
>On Mon, 1 Feb 1999, Seymour Glass wrote:
>
>> Why is this behavior correct? Since
>>
>> 10 ^. 1208021x
>> |domain error
>>
>> I would have thought that 10 ([: >: <.@^.)1208021x
>> would fail too. Why doesn't it? It looks like the
>> 1208021x is being converted to 1208021, but I don't
>> see why.
>>
>> Henry Rich
>>
>> -----Original Message-----
>> From: Andrew Nikitin <nsg@voxel.kharkov.ua>
>> To: forum@JSoftware.Com <forum@JSoftware.Com>
>> Date: Monday, February 01, 1999 11:39
>> Subject: Jforum: @: and [:
>>
>>
>> >
>> > 10 ([: >: <.@^.)1208021x
>> >7
>> > 10 (>: @: <.@^.)1208021x
>> >|domain error
>> >| 10 (>:@:<.@^.)1208021
>> >
>> >After some meditation i have decided that such behaviour is completely
>> >correct, but it still looks some puzzling.
>> >
>> >BTW, this example shows that composition in J isn't associative as it is
in
>> >mathematic, so, be careful.
>> >
>> >
>> >nsg
>> >
>> >
>> >
>>
>---------------------------------------------------------------------------
>> -----
>> >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
>>
>
>
>---------------------------------------------------------------------------
-----
>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 Tue Feb 2 09:32:16 1999
From: "Nichols, Peter" <pnichols@sprinc.com>
Subject: Jforum: Amending columns in a matrix
Date: Fri, 29 Jan 1999 15:25:17 -0600
charset="iso-8859-1"
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by lists.interlog.com id JAA23049
Given a simple matrix
a=:i.3 3
a
0 1 2
3 4 5
6 7 8
I can get the second column.
1{"1 a
1 4 7
But if I try to amend it
in an intuitive way.
(12 13 14) 1}"1 a
�rank error
� (12 13 14) 1}"1 a
What is the J equivalent of APL's
a[;1]assign iota 3
Thanks
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 10:19:49 1999
Date: Tue, 02 Feb 1999 10:00:00 -0500
From: Cliff Reiter <reiterc@lafvax.lafayette.edu>
Subject: Re: Jforum: Amending columns in a matrix
Organization: Lafayette College
Content-type: text/plain; charset=iso-8859-1
References: <01074BD4EEC4D1118E3200805F6542A7177DAE@SPRDALLAS2.lafayette.edu>
A solution to your question using rank is
(read it as amend position 1 in each row)
12 13 14 (1)}"0 1 i.3 3
0 12 2
3 13 5
6 14 8
More generally
12 13 14(<(<i.0);1)} i. 3 3
0 12 2
3 13 5
6 14 8
Studying the boxed cases of m. in "from" is worthwhile
Best,
Cliff
Nichols, Peter wrote:
>
> Given a simple matrix
>
> a=:i.3 3
> a
> 0 1 2
> 3 4 5
> 6 7 8
>
> I can get the second column.
>
> 1{"1 a
> 1 4 7
>
> But if I try to amend it
> in an intuitive way.
>
> (12 13 14) 1}"1 a
> �rank error
> � (12 13 14) 1}"1 a
>
> What is the J equivalent of APL's
> a[;1]assign iota 3
--
Clifford A. Reiter
Mathematics Department, Lafayette College
Easton, PA 18042 USA, 610-330-5277
^^^ Note new phone exchange
http://www.lafayette.edu/~reiterc
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 10:24:07 1999
From: mellemf@nimo.com
X-Lotus-FromDomain: NMPC
Date: Tue, 2 Feb 1999 09:58:07 -0500
Subject: Re: Jforum: Amending columns in a matrix
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline
a=:i.3 3
(12 13 14) 1}"0 1 a
Your intuition is good , but apply it (with rank ) to both sides.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 10:32:24 1999
Date: Tue, 2 Feb 1999 16:17:19 +0100 (MET)
From: Martin Neitzel <neitzel@gaertner.de>
Subject: Re: Jforum: Amending columns in a matrix
(12 13 14) 1}"0 1 a
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 14:47:52 1999
Delivered-To: fixup-forum@jsoftware.com@fixme
Date: Tue, 02 Feb 1999 09:43:14 -0800
From: greg heil <gheil@uswest.net>
Subject: Re: Jforum: Amending columns in a matrix
References: <01074BD4EEC4D1118E3200805F6542A7177DAE@SPRDALLAS2>
"Nichols, Peter" wrote:
> What is the J equivalent of APL's
a[;1]assign iota 3
try
(i.3) 1}"0 1 a=:i.3 3
0 0 2
3 1 5
6 2 8
or
(9) 1}"1 a
0 9 2
3 9 5
6 9 8
PS your clock seems to be off by 3 days, its Ground Hogs
day here in Seattle (nobody is seeing their shadow either;-)
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 Tue Feb 2 16:39:32 1999
From: "J Tibollo" <jtibollo@backassociates.com>
Subject: Re: Jforum: Amending columns in a matrix
Date: Tue, 2 Feb 1999 16:30:07 -0500
charset="iso-8859-1"
Can someone offer a technical explanation of why:
(i.3) 1}"0 1 a=:i.3 3
gives:
0 0 2
3 1 5
6 2 8
and,
a=:i.3 3
1 {"1 a
1 4 7
while,
(i.3) 1 }"1 a
|rank error
| (i.3) 1}"1 a
Just curious - why do you need to use "0 1 instead of "1 ?
Thanks,
Joe Tibollo
-----Original Message-----
From: greg heil <gheil@uswest.net>
Date: Tuesday, February 02, 1999 2:38 PM
Subject: Re: Jforum: Amending columns in a matrix
>"Nichols, Peter" wrote:
>
>> What is the J equivalent of APL's
> a[;1]assign iota 3
>
>try
> (i.3) 1}"0 1 a=:i.3 3
>0 0 2
>3 1 5
>6 2 8
>
>or
> (9) 1}"1 a
>0 9 2
>3 9 5
>6 9 8
>
>PS your clock seems to be off by 3 days, its Ground Hogs
>day here in Seattle (nobody is seeing their shadow either;-)
>
>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
>
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 16:43:40 1999
From: "Nichols, Peter" <pnichols@sprinc.com>
Subject: Jforum: Reassigning a matrix in a column
Date: Mon, 1 Feb 1999 09:40:40 -0600
If a=:i.3 3
How do I make the 2nd column be 98 99 100?
Thanks
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 17:15:48 1999
Date: Tue, 02 Feb 1999 16:03:13 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Amending columns in a matrix
References: <000a01be4ef3$379e6720$91d23dcf@billy>
I will try my best, which is not much...
In the help facility you can find in Definitions:
The three ranks (in the order monadic, left, and right) are also indicated, using
the symbol _ for an infinite (unbounded) rank, and with ranks dependent on the ranks
of argument verbs shown as mu, lv, etc.
So, if you do:
b=: 10 40 70
a=:i.3 3
b 1}"0 1 a
Does what you know now try this:
b=:1 3$10 40 70
then:
b 1}"0 2 a
b 1}"1 0 2 a
b 1}1 2 a
b 1}1 1 2 a
Regards/Paul
J Tibollo wrote:
> Can someone offer a technical explanation of why:
>
> (i.3) 1}"0 1 a=:i.3 3
>
> gives:
>
> 0 0 2
> 3 1 5
> 6 2 8
>
> and,
>
> a=:i.3 3
>
> 1 {"1 a
> 1 4 7
>
> while,
>
> (i.3) 1 }"1 a
> |rank error
> | (i.3) 1}"1 a
>
> Just curious - why do you need to use "0 1 instead of "1 ?
>
> Thanks,
> Joe Tibollo
>
> -----Original Message-----
> From: greg heil <gheil@uswest.net>
> To: forum@jsoftware.com <forum@jsoftware.com>
> Date: Tuesday, February 02, 1999 2:38 PM
> Subject: Re: Jforum: Amending columns in a matrix
>
> >"Nichols, Peter" wrote:
> >
> >> What is the J equivalent of APL's
> > a[;1]assign iota 3
> >
> >try
> > (i.3) 1}"0 1 a=:i.3 3
> >0 0 2
> >3 1 5
> >6 2 8
> >
> >or
> > (9) 1}"1 a
> >0 9 2
> >3 9 5
> >6 9 8
> >
> >PS your clock seems to be off by 3 days, its Ground Hogs
> >day here in Seattle (nobody is seeing their shadow either;-)
> >
> >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
> >
>
> --------------------------------------------------------------------------------
> 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 Tue Feb 2 17:21:46 1999
Date: Tue, 02 Feb 1999 16:07:27 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Reassigning a matrix in a column
References: <01074BD4EEC4D1118E3200805F6542A7177DB2@SPRDALLAS2>
a=:i.3 3
b=:98 99 100
b 1}"0 1 a
98 99 100(1)}"0 1 a
"Nichols, Peter" wrote:
> If a=:i.3 3
> How do I make the 2nd column be 98 99 100?
>
> Thanks
>
> --------------------------------------------------------------------------------
> 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 Tue Feb 2 17:24:34 1999
Date: Tue, 2 Feb 1999 23:12:27 +0100 (MET)
From: Martin Neitzel <neitzel@gaertner.de>
Subject: Re: Jforum: Amending columns in a matrix
> Can someone offer a technical explanation of why:
>
> [some stuff elided]
>
> 1 {"1 a
> 1 4 7
>
> while,
>
> (i.3) 1 }"1 a
> |rank error
> | (i.3) 1}"1 a
>
> Just curious - why do you need to use "0 1 instead of "1 ?
First of all, { and } are different beasts. While 1{ looks
very similar to 1} , the structure is completely different.
1 { a left argument next to a dyad (no valid phrase as is)
1 } adverb (rather, "adnoun") next to its operand "1" => a verb
Next, while { and } look symmetric and they are related, they have
nevertheless the own requirements and argument ranks.
Now to your question:
> Just curious - why do you need to use "0 1 instead of "1 ?
(i.3) 1 }"1 a
^^^^^ ^ NB. hope you have a monospaced font
is
x 1 }"1 y
^^^
is
x verb"1 y
Now, verb"1 (a new verb on its own) is clearly applied dyadically.
In the dyadic case, verb"1 is just a short form for verb"1 1 .
(Which again is just a short form for verb"1 1 1 with order 'mr lr rr';
that's where the 3{body}amp;.|.n kicks in in the Dictionary's u"n entry.)
But verb"0 1 is what you want: mate scalars from the left with
rows from the right; for each pair, do the 1} shove-it-in thing.
(1}) "1 1 yields a rank error because a row (0 1 2) would be
paired with any other row from the left (say, 4 5 6). But
0 1 2 (1}) 4 5 6
rightly complains at you because you cannot amend the scalar 5 with
an entire vector. The Dictionary requires for x m} y :
$x must be a suffix of $m{y
Hope this helps,
Martin
PS: 1000 thanks for omitting that HTML encoding this time!
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 18:40:57 1999
Date: Tue, 02 Feb 1999 17:30:32 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Reassigning a matrix in a column
References: <01074BD4EEC4D1118E3200805F6542A7177DB2@SPRDALLAS2> <36B7771F.984D21@sympatico.ca>
Some timing issues.
Now, please do not forget that internally, data is stored in a ravel wise order and
therefore:
100000 (6!:2)'b 1}"0 1 a'
0.0001395
100000 (6!:2)'|:b 1}|:a'
0.000106
So, if two transposes keep the timing faster than directly replacing a column, what
about doing a transpose at the start of your procedure, replace all the rows you need
to replace then transpose again...
Regards/Paul
Paul Gauthier wrote:
> a=:i.3 3
> b=:98 99 100
> b 1}"0 1 a
> 98 99 100(1)}"0 1 a
>
> "Nichols, Peter" wrote:
>
> > If a=:i.3 3
> > How do I make the 2nd column be 98 99 100?
> >
> > Thanks
> >
> > --------------------------------------------------------------------------------
> > 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
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 18:45:01 1999
From: "Nollaig MacKenzie" <Nollaig@YorkU.CA>
Date: Tue, 02 Feb 1999 18:30:41 -0500 (EST)
In-Reply-To: <01074BD4EEC4D1118E3200805F6542A7177DAE@SPRDALLAS2>
Subject: Re: Jforum: Amending columns in a matrix
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by lists.interlog.com id SAA23826
On Fri, 29 Jan 1999 15:25:17 -0600, Nichols, Peter wrote:
>Given a simple matrix
>
> a=:i.3 3
> a
>0 1 2
>3 4 5
>6 7 8
>
>
>I can get the second column.
>
> 1{"1 a
>1 4 7
>
>But if I try to amend it
>in an intuitive way.
>
> (12 13 14) 1}"1 a
> rank error
> (12 13 14) 1}"1 a
>
>What is the J equivalent of APL's
>a[;1]assign iota 3
>
This kicked in a motto I made for myself the last time I was writing
in J (3.05): If the cell-rank don't work, try the frame.
Consider:
a=. i. 3 3
b=. i.3 3 3
v=. 11 14 17
m=. 100*i.3 3
v 1 }"1 a
�rank error
� v 1}"1 a
v 1}"_1 a
0 11 2
3 14 5
6 17 8
m 1}"1 b
�rank error
� m 1}"1 b
m 1}"_2 b
0 0 2
3 100 5
6 200 8
9 300 11
12 400 14
15 500 17
18 600 20
21 700 23
24 800 26
m 1}"2 b
�rank error
� m 1}"2 b
m 1}"_1 b
0 1 2
0 100 200
6 7 8
9 10 11
300 400 500
15 16 17
18 19 20
600 700 800
24 25 26
I never understood this, since frame-rank (negative n in "n) and
cell-rank (positive n in "n) should behave something like this,
shouldn't they?:
If the rank of A is R, and m+n = R, then "(-m) is equivalent to "n
eh?
Cheers, N.
--
hostname if connected: Amhuinnsuidhe.DynIP.COM
http://www.yorku.ca/faculty/academic/nollaig/
For PGP public key, send e-mail to: key@Four11.com
with body: Nollaig@YorkU.CA
--
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 19:13:40 1999
Date: Wed, 3 Feb 1999 01:03:53 +0100 (MET)
From: Martin Neitzel <neitzel@gaertner.de>
Subject: Re: Jforum: Amending columns in a matrix
> I never understood this, since frame-rank (negative n in "n) and
> cell-rank (positive n in "n) should behave something like this,
> shouldn't they?:
>
> If the rank of A is R, and m+n = R, then "(-m) is equivalent to "n
>
> eh?
Yes if you consider _one_ spec m with respect to _one_ argument A.
Not if you consider _one_ sepc m with respect to _two_ args A and B.
Then:
> a=. i. 3 3
> b=. i.3 3 3
> v=. 11 14 17
> m=. 100*i.3 3
> v 1 }"1 a => "1 1
> v 1}"_1 a => "_1 _1 (here positively: "0 1)
> m 1}"1 b => "1 1
> m 1}"2 b => "2 2
> m 1}"_1 b => "_1 _1 (here positively: "1 1)
Do you see the differences now? After you duplicate the single
rank spec properly for the left and rigth side, you can do
the usual frame/cell arithmetic. One each side separately.
Martin
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 19:34:21 1999
Date: Wed, 3 Feb 1999 01:22:18 +0100 (MET)
From: Martin Neitzel <neitzel@gaertner.de>
Subject: Re: Jforum: Amending columns in a matrix
Oops, I trimmed a bit too much. So here it comes:
> m 1}"_2 b => "_2 _2 (here positively: "0 0)
OK OK... it's not strictly "positively". "Cell-wise", then.
Martin
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 20:56:34 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: Amending columns in a matrix
Date: Tue, 2 Feb 1999 20:46:10 -0500
-----Original Message-----
From: Martin Neitzel <neitzel@gaertner.de>
Date: Tuesday, February 02, 1999 19:03
Subject: Re: Jforum: Amending columns in a matrix
>> I never understood this, since frame-rank (negative n in "n) and
>> cell-rank (positive n in "n) should behave something like this,
>> shouldn't they?:
>>
>> If the rank of A is R, and m+n = R, then "(-m) is equivalent to "n
>>
>> eh?
>
>Yes if you consider _one_ spec m with respect to _one_ argument A.
>Not if you consider _one_ sepc m with respect to _two_ args A and B.
[ useful and accurate examples omitted ]
But advanced users should note that "(-m) is actually equivalent
to "n"_ rather than to "n (in other words, the derived verb has infinite
rank):
0:@(]"_1) i. 5 5
0
0:@(]"1) i. 5 5
0 0 0 0 0
0:@(]"1"_) i. 5 5
0
or more plainly:
(]"1) b. 0
1 1 1
(]"_1) b. 0
_ _ _
Henry Rich
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 2 22:30:18 1999
Delivered-To: fixup-forum@jsoftware.com@fixme
Date: Tue, 02 Feb 1999 19:20:14 -0800
From: greg heil <gheil@uswest.net>
Subject: Re: Jforum: Amending columns in a matrix
References: <000a01be4ef3$379e6720$91d23dcf@billy>
J Tibollo wrote:
> Can someone offer a technical explanation of why:
>... <snip> while,
> (i.3) 1 }"1 a
> |rank error
> | (i.3) 1}"1 a
Here you are trying to cram (technical term there;-) a vector
(i.3) into the second (1}) scalar of each row ("1).
> Just curious - why do you need to use "0 1 instead of "1 ?
The 0 says to use scalars as the left argument of the amend
action and the 1 to use row vectors as the, corresponding,
right argument. To just use one 1 implies cramming (getting
technical again;) a vector into a scalar slot as that 1 is
repeated to derive the verbs rank for both left and right args.
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 Feb 3 10:57:14 1999
From: "Ondrus, Milan" <milan.ondrus@csfb.com>
Subject: Jforum: fmt
Date: Wed, 3 Feb 1999 16:38:43 +0100
I'm looking at the utility fmt in the script <format>
'b4' fmt 3456 gives the char-matrix 3456
but
'z4' fmt 3456 gives the char-matrix 0456 ... inserts a zero at the first
position and wipes out the 3.
'z3' fmt 345 gives 045 ... again the first position disappears.
This z-formatting seems to behave inconsistent with b-formatting or do I
see that wrong?
.....................
/Milan Ondrus
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Wed Feb 3 12:55:36 1999
Date: Wed, 03 Feb 1999 17:11:49 -0800
From: Roger Stokes <qk57@dial.pipex.com>
Subject: Jforum: A Book About J
References: <89FE544F2452D21190A50060B0684BF309B0AB@szrh00305.tszrh.csfb.com>
If anyone would like to take a look at
http://dspace.dial.pipex.com/rstokes/book.htm
you will see some chapters of a book about J
which I'm in the middle of writing.
About one-third of the projected text is
available on the website, and and another one-third
is in draft, to appear soon.
Your comments and criticisms would be very much appreciated.
Many thanks.
--
Regards: Roger Stokes rstokes@dial.pipex.com
- or -
qk57@dial.pipex.com
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 13:20:09 1999
Date: 4 Feb 99 10:01:59 -0800
Subject: Re: Jforum: Amending columns in a matrix
From: "Piet de Jong" <piet.dejong@commerce.ubc.ca>
X-Fontfamily: Geneva
X-Fontsize: 12
Content-Transfer-Encoding: quoted-printable
On the subject of ammending colums, what is the easiest way to
ammend a single row of a matrix. By this I mean an adverb (call it A
say) such that
( 5 1 2) 1 A i.3 3
puts the row 5 1 2 in the second (ie 1) row of i.3 3.
I realize you can transpose the matrix do the column amendment as
discussed here and transpose again. But this seems clumsy. Is there
an easier, more
>>transparent<<<< way.
This also leads me to the question why ammend is "atom" rather than
"item" based. Most of the time J (as I understand it) is "item"
based.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 13:57:21 1999
Date: Thu, 4 Feb 1999 20:33:55 +0200 (EET)
From: Oleg Kobchenko <gccinc@usa.net>
Subject: Re: Jforum: Amending columns in a matrix
In-Reply-To: <B2DF2098-985F40@137.82.66.177>
M=: i. 3 3
M
0 1 2
3 4 5
6 7 8
_1 _2 _3 (0)} M
_1 _2 _3
3 4 5
6 7 8
In J 'items' ARE the rows, etc, generally the top
of the rank.
On 4 Feb 1999, Piet de Jong wrote:
> On the subject of ammending colums, what is the easiest way to
>
> ammend a single row of a matrix. By this I mean an adverb (call it A
> say) such that
>
>
> ( 5 1 2) 1 A i.3 3
>
>
> puts the row 5 1 2 in the second (ie 1) row of i.3 3.
>
>
> I realize you can transpose the matrix do the column amendment as
> discussed here and transpose again. But this seems clumsy. Is there
> an easier, more
>
> >>transparent<<<< way.
>
>
> This also leads me to the question why ammend is "atom" rather than
> "item" based. Most of the time J (as I understand it) is "item"
> based.
>
> --------------------------------------------------------------------------------
> 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 Thu Feb 4 13:57:53 1999
Date: Thu, 4 Feb 1999 19:44:48 +0100 (MET)
From: Martin Neitzel <neitzel@gaertner.de>
Subject: Re: Jforum: Amending columns in a matrix
Piet de Jong:
>
> On the subject of ammending colums, what is the easiest way to
> ammend a single row of a matrix. By this I mean an adverb (call it A
> say) such that
>
> ( 5 1 2) 1 A i.3 3
>
> puts the row 5 1 2 in the second (ie 1) row of i.3 3.
Well, this one is easy. Your A _is_ amend (}).
> This also leads me to the question why ammend is "atom" rather than
> "item" based. Most of the time J (as I understand it) is "item"
> based.
x m} y _is_ item based for the most trivial case (an open m).
The value(s) in m will then indicate which ITEMs of y will be amended.
It's only when m is boxed that other axes than the first axis come
into play, i.e. finer selections of y then items.
So, where do you get the impression that m} would be atom based?
All in all, I have the impression there must be some misunderstanding.
Martin
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 14:03:50 1999
Date: Thu, 04 Feb 1999 12:50:17 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Amending columns in a matrix
References: <B2DF2098-985F40@137.82.66.177>
You just want the standard amand adverb...
(5 1 2)1}i.3 3
5 1 2(1)}i.3 3
N.B.: two transposes are not clumsy if you do it for efficiency
Regards/Paul
Piet de Jong wrote:
> On the subject of ammending colums, what is the easiest way to
> ammend a single row of a matrix. By this I mean an adverb (call it A say) such that
>
> ( 5 1 2) 1 A i.3 3
>
> puts the row 5 1 2 in the second (ie 1) row of i.3 3.
>
> I realize you can transpose the matrix do the column amendment as discussed here and transpose again. But this seems clumsy. Is there an easier, more
> >>transparent<< way.
>
> This also leads me to the question why ammend is "atom" rather than "item" based. Most of the time J (as I understand it) is "item" based.
> -------------------------------------------------------------------------------- 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 Thu Feb 4 15:23:43 1999
Date: 4 Feb 99 12:07:10 -0800
Subject: Re: Jforum: Amending columns in a matrix
From: "Piet de Jong" <piet.dejong@commerce.ubc.ca>
X-Fontfamily: Geneva
X-Fontsize: 12
Content-Transfer-Encoding: quoted-printable
On the subject of the amend verb , here is someting that I sometimes
wish for in J.
Consider u} where u is a verb. Then u "sees " the argument on the
left. However this is not the case with the adverb \. Thus with u\,
the verb u does not "see "the argument on the left. Sometimes it
would be nice to have u be able to take both left and right arguments.
For example
12 (+/%#) \ x
computes the mean of each block ("infixes") of 12 items in x.
However if you want a weighted average with weights specified on the
left then things are not so convenient. You can do it but its a bit
of workaround. Ideally you would want u to be able determine the
length and order of the weighted average from the left and apply them
to the right. Can this be done easily and >>transparently<<<<.
(No explicit definitions please)
This brings me to another issue which I have not figured out in J.
Why and when should I prefer an explicit definition to a tacit one.?
Is it right that generally speaking tacit ones are much preferred (the
dictionary etc seems to intimate as much).
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 15:57:47 1999
Date: Thu, 4 Feb 1999 22:38:29 +0200 (EET)
From: Oleg Kobchenko <gccinc@usa.net>
Subject: Re: Jforum: Amending columns in a matrix
In-Reply-To: <B2DF3DEF-9F44A8@137.82.66.177>
It's not very clear what exactly you were looking for.
One of my supositions would be:
[x=. 3 4 5 6
3 4 5 6
NB. simple average of infixes
2 (+/ % #)\ x
3.5 4.5 5.5
NB. weighted average of infixes
0.7 0.3 ([: (+/ % #)"1 [ *"1 #@[ ]\ ]) x
1.65 2.15 2.65
On 4 Feb 1999, Piet de Jong wrote:
> On the subject of the amend verb , here is someting that I sometimes
> wish for in J.
>
>
> Consider u} where u is a verb. Then u "sees " the argument on the
> left. However this is not the case with the adverb \. Thus with u\,
> the verb u does not "see "the argument on the left. Sometimes it
> would be nice to have u be able to take both left and right arguments.
>
>
> For example
>
>
> 12 (+/%#) \ x
>
>
> computes the mean of each block ("infixes") of 12 items in x.
> However if you want a weighted average with weights specified on the
> left then things are not so convenient. You can do it but its a bit
> of workaround. Ideally you would want u to be able determine the
> length and order of the weighted average from the left and apply them
> to the right. Can this be done easily and >>transparently<<<<.
> (No explicit definitions please)
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 16:05:09 1999
Date: 4 Feb 99 12:47:55 -0800
Subject: Re: Jforum: Amending columns in a matrix
From: "Piet de Jong" <piet.dejong@commerce.ubc.ca>
X-Fontfamily: Geneva
X-Fontsize: 12
Content-Transfer-Encoding: quoted-printable
Thanks to all for the information that amend IS item based.
The confusion arose from the version of J 6.2 (1992) I run on my
HP200LX. In this version amend is atom based. and indeed a statement
such as
(5 1 2) 1} i. 3 3
yields a "rank error". Also the dictionary I use
(the most studied book in my library) , relating to version 6.2,
appears to state that amend is atom based.
On my newer version of J, (3.05 or thereabouts) amend indeed behaves
as item based. I'm glad it is item based since this seems to make
more sense.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 16:23:07 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: Amending columns in a matrix
Date: Thu, 4 Feb 1999 16:07:09 -0500
boundary="----=_NextPart_000_00B5_01BE5058.6AC51E80"
This is a multi-part message in MIME format.
------=_NextPart_000_00B5_01BE5058.6AC51E80
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
I think it's impossible to do this implicitly. But explicitly you could =
write:
NB. Adverb: y. is anything, x. is anything, (x.&u.) is applied to each =
infix of length #x.
appliedtoinfixes =3D: 1 : 0
[: y.
:
(#x.) (x.&u.)\ y.
)
and then:
wsum =3D. +/ @: *
1 2 3 wsum appliedtoinfixes i. 10
8 14 20 26 32 38 44 50
the monadic case is defined so that:
wsum appliedtoinfixes i. 10
|valence error
| [:y.
When to use tacit vs. explicit is a personal choice. I use tacits for =
functions whose
definition is not likely to change, i. e. my utility functions, and =
explicits for
application-specific work. And, as here, explicits when you need to =
pass x. or y.
into an adverb or conjunction.
Henry Rich
-----Original Message-----
From: Piet de Jong <piet.dejong@commerce.ubc.ca>
Date: Thursday, February 04, 1999 15:07
Subject: Re: Jforum: Amending columns in a matrix
On the subject of the amend verb , here is someting that I sometimes =
wish for in J.
=20
Consider u} where u is a verb. Then u "sees " the argument on the =
left. However this is not the case with the adverb \. Thus with u\, the =
verb u does not "see "the argument on the left. Sometimes it would be =
nice to have u be able to take both left and right arguments.
=20
For example=20
=20
12 (+/%#) \ x
=20
computes the mean of each block ("infixes") of 12 items in x. =
However if you want a weighted average with weights specified on the =
left then things are not so convenient. You can do it but its a bit of =
workaround. Ideally you would want u to be able determine the length and =
order of the weighted average from the left and apply them to the right. =
Can this be done easily and >>transparently<<. (No explicit definitions =
please)=20
=20
This brings me to another issue which I have not figured out in J. =
Why and when should I prefer an explicit definition to a tacit one.? Is =
it right that generally speaking tacit ones are much preferred (the =
dictionary etc seems to intimate as much).=20
=20
=20
=20
=
-------------------------------------------------------------------------=
------- J Forum: for information about this list, see =
http://www.jsoftware.com/forum.htm
------=_NextPart_000_00B5_01BE5058.6AC51E80
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">
<HTML>
<HEAD>
<META content=3Dtext/html;charset=3Diso-8859-1 =
http-equiv=3DContent-Type>
<META content=3D'"MSHTML 4.72.3509.100"' name=3DGENERATOR>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT color=3D#000000><FONT size=3D3>I think it's impossible to do =
this=20
implicitly. But explicitly you could write:</FONT></FONT><FONT=20
size=3D3></FONT></DIV>
<DIV><FONT color=3D#000000><FONT size=3D3></FONT></FONT><FONT=20
size=3D3></FONT> </DIV>
<DIV><FONT face=3D"" size=3D3>NB. Adverb: y. is anything, x. is =
anything,=20
(x.&u.) is applied to each infix of length #x.<BR>appliedtoinfixes =
=3D: 1 :=20
0<BR>[: y.<BR>:<BR>(#x.) (x.&u.)\ y.<BR>)<BR></FONT></DIV>
<DIV>and then:</DIV>
<DIV> </DIV>
<DIV> wsum =3D. +/ @: *<BR> 1 2 3 wsum =
appliedtoinfixes i.=20
10<BR>8 14 20 26 32 38 44 50</DIV>
<DIV> </DIV>
<DIV>the monadic case is defined so that:</DIV>
<DIV><BR> wsum appliedtoinfixes i. 10<BR>|valence=20
error<BR>| [:y.<BR></DIV>
<DIV>When to use tacit vs. explicit is a personal choice. I use =
tacits for=20
functions whose</DIV>
<DIV>definition is not likely to change, i. e. my utility functions, and =
explicits for</DIV>
<DIV>application-specific work. And, as here, explicits when you =
need to=20
pass x. or y.</DIV>
<DIV>into an adverb or conjunction.</DIV>
<DIV> </DIV>
<DIV>Henry Rich</DIV>
<DIV><FONT face=3DArial size=3D2><B></B></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2><B>-----Original =
Message-----</B><BR><B>From:=20
</B>Piet de Jong <<A=20
href=3D"mailto:piet.dejong@commerce.ubc.ca">piet.dejong@commerce.ubc.ca</=
A>><BR><B>To:=20
</B><A href=3D"mailto:forum@jsoftware.com">forum@jsoftware.com</A> =
<<A=20
href=3D"mailto:forum@jsoftware.com">forum@jsoftware.com</A>><BR><B>Dat=
e:=20
</B>Thursday, February 04, 1999 15:07<BR><B>Subject: </B>Re: Jforum: =
Amending=20
columns in a matrix<BR><BR></DIV>
<BLOCKQUOTE=20
style=3D"BORDER-LEFT: #000000 solid 2px; MARGIN-LEFT: 5px; PADDING-LEFT: =
5px"></FONT>On=20
the subject of the amend verb , here is someting that I sometimes =
wish for=20
in J.<BR><BR>Consider u} where u is a verb. Then u "sees " =
the=20
argument on the left. However this is not the case with the adverb =
\. Thus=20
with u\, the verb u does not "see "the argument on the =
left.=20
Sometimes it would be nice to have u be able to take both left and =
right=20
arguments.<BR><BR>For example <BR><BR>12 (+/%#) \ x<BR><BR>computes =
the mean=20
of each block ("infixes") of 12 items in x. However if you =
want a=20
weighted average with weights specified on the left then things are =
not so=20
convenient. You can do it but its a bit of workaround. Ideally you =
would=20
want u to be able determine the length and order of the weighted =
average=20
from the left and apply them to the right. Can this be done easily =
and=20
>>transparently<<. (No explicit definitions please) =
<BR><BR>This=20
brings me to another issue which I have not figured out in J. Why =
and when=20
should I prefer an explicit definition to a tacit one.? Is it right =
that=20
generally speaking tacit ones are much preferred (the dictionary etc =
seems=20
to intimate as much).=20
=
<BR><BR><BR><BR>---------------------------------------------------------=
-----------------------=20
J Forum: for information about this list, see=20
http://www.jsoftware.com/forum.htm</BLOCKQUOTE></BODY></HTML>
------=_NextPart_000_00B5_01BE5058.6AC51E80--
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 16:50:20 1999
From: "Nichols, Peter" <pnichols@sprinc.com>
Subject: RE: Jforum: Reassigning a matrix in a column
Date: Tue, 2 Feb 1999 15:47:54 -0600
charset="iso-8859-1"
Never mind. For some reason my posts are going out several days after I
actually send them. I've already gotten several great answers.
-----Original Message-----
From: Nichols, Peter [SMTP:pnichols@sprinc.com]
Sent: Monday, February 01, 1999 9:41 AM
To: 'forum@jsoftware.com'
Subject: Jforum: Reassigning a matrix in a column
If a=:i.3 3
How do I make the 2nd column be 98 99 100?
Thanks
----------------------------------------------------------------------------
----
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 Thu Feb 4 17:47:25 1999
From: "Nichols, Peter" <pnichols@sprinc.com>
Subject: Jforum: Boxing Columns
Date: Wed, 3 Feb 1999 14:02:24 -0600
I got this function from Roger Stokes book in progress. It boxes the
x-cells of y.
cells =: 4 : '< " x. y.'
In a 3x3 matrix, the 0-cells are the individual cells. 1-cells are the
rows, and 2-cells are the whole matrix. But how would you box the columns
of a matrix? I know I could transpose but it seems like there should be a
more elegant way.
Thanks.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 18:49:10 1999
From: "John D. Baker" <bakerjd@kos.net>
Subject: Jforum: Tacit and Explicit
Date: Thu, 4 Feb 1999 18:35:57 -0500
charset="iso-8859-1"
Piet de Jong writes:
>This brings me to another issue which I have not figured out in J. Why and
when should I prefer an explicit >definition to a tacit one.? Is it right
that generally speaking tacit ones are much preferred (the dictionary etc
>seems to intimate as much).
Sometime after I first learned J I realized that spending inordinate amounts
of
time looking for tacit solutions to problems that are trivially solved with
explicit
expressions is J's variation on APL's one-liner-real-men-don't-loop disease.
The decision to use tacit vs. explicit definitions should be based on
rational
criteria. Here's my J checklist. I use similar guidelines when
programming
in APL and even when mulling over the macro/function dilemma in C.
1) If you can quickly bash off a tacit definition without stopping to think
about it use a tacit expression.
2) If your verb is the argument of an adverb or conjunction and is going
to be executed thousands (or more) times, e,g.
u&.> somebighonkinglist
It often, (but not always), pays to use tacit definitions. If a line is
going
to be executed only once the tacit/explicit issue is moot. Your time
is better spent thinking about what's on that line rather than looking
for
a tacit version of it.
3) If you are creating local words inside other definitions try and make
them
tacit. Keep in mind that excessive embedding is pretty clear sign that
you need
a locale or class object.
4) If your word naturally occupies a few lines of code and all of the above
applies use an explicit definition. The first time you have to turn
on the debugger you will be rewarded for your tacit restraint.
5) If you are interested in exploring J semantics solely for it's own sake
then tacits are superb teachers. It is for this reason that tacits are
so frequently used in The J Dictionary.
6) If you have a lot of time on your hands and you belong to some
bizarre ultra-loony-mad-programmer's-cult, (like
the dwindling but still in denial pure Java sect), then I'm sorry --
you can't use explicit
definitions because the J interpreter detects fanatics and
automatically disables explicit definitions.
John D. Baker
bakerjd@kos.net
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Thu Feb 4 21:30:30 1999
Date: Fri, 05 Feb 1999 03:18:46 +0100
From: "d.alis" <dalis@balcab.ch>
Subject: Re: Jforum: Boxing Columns
References: <01074BD4EEC4D1118E3200805F6542A7177DBA@SPRDALLAS2>
Whatever J is about, surely it is about writing
cells =: <"
instead of
cells =: 4 : '< " x. y.'
Or does no-one agree?
DAL
Nichols, Peter wrote:
> I got this function from Roger Stokes book in progress. It boxes the
> x-cells of y.
>
> cells =: 4 : '< " x. y.'
>
> In a 3x3 matrix, the 0-cells are the individual cells. 1-cells are the
> rows, and 2-cells are the whole matrix. But how would you box the columns
> of a matrix? I know I could transpose but it seems like there should be a
> more elegant way.
>
> Thanks.
>
> --------------------------------------------------------------------------------
> 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 Thu Feb 4 21:47:49 1999
Date: Fri, 05 Feb 1999 03:27:44 +0100
From: "d.alis" <dalis@balcab.ch>
Subject: Re: Jforum: Amending columns in a matrix
References: <B2DF477D-A18397@137.82.66.177>
The Dictionary needs to be read with care.
Compare the noun and verb cases...
a=: 1 3 5
Anoun =. a
Averb =. a"_
100 200 300 Anoun } i. 6 3
0 1 2
100 200 300
6 7 8
100 200 300
12 13 14
100 200 300
100 200 300 Averb } i. 6 3
0 100 2
200 4 300
6 7 8
9 10 11
12 13 14
15 16 17
Piet de Jong wrote:
> Thanks to all for the information that amend IS item based.
>
> The confusion arose from the version of J 6.2 (1992) I run on my
> HP200LX. In this version amend is atom based. and indeed a statement such as
>
> (5 1 2) 1} i. 3 3
>
> yields a "rank error". Also the dictionary I use
> (the most studied book in my library) , relating to version 6.2, appears to state that amend is atom based.
>
> On my newer version of J, (3.05 or thereabouts) amend indeed behaves as item based. I'm glad it is item based since this seems to make more sense.
>
> -------------------------------------------------------------------------------- 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 Fri Feb 5 04:11:00 1999
References: <89FE544F2452D21190A50060B0684BF309B0AB@szrh00305.tszrh.csfb.com>
Organization: Voxel
From: "Andrew Nikitin" <nsg@voxel.kharkov.ua>
Date: Thu, 4 Feb 1999 13:52:38 +0200 (EET)
Subject: Re: Jforum: fmt
3-Feb-99 16:38 Ondrus, Milan wrote:
> I'm looking at the utility fmt in the script <format>
Conjunction expandby marked by label 'v' instead of 'c' in scriptdoc.
Line 3 in hexdump verb would be better changed from
bar=. 179{a. NB. original version
into something like
bar=. 9{,9!:6 ''
nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 5 04:12:09 1999
Date: Fri, 05 Feb 1999 02:50:51 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Boxing Columns
References: <01074BD4EEC4D1118E3200805F6542A7177DBA@SPRDALLAS2>
Please read the Phrase book Chapter 5 on structures, section C on special matrices
and lists.
You will find:
m42=:<"1@|: NB.Box each column of a matrix
Yes, it is tranposed...Nope I don't know of a more elegant way...
"Nichols, Peter" wrote:
> I got this function from Roger Stokes book in progress. It boxes the
> x-cells of y.
>
> cells =: 4 : '< " x. y.'
>
> In a 3x3 matrix, the 0-cells are the individual cells. 1-cells are the
> rows, and 2-cells are the whole matrix. But how would you box the columns
> of a matrix? I know I could transpose but it seems like there should be a
> more elegant way.
>
> Thanks.
>
> --------------------------------------------------------------------------------
> 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 Fri Feb 5 07:48:32 1999
References: <36BAB0EA.18772AE@sympatico.ca>
Organization: Voxel
From: "Andrew Nikitin" <nsg@voxel.kharkov.ua>
Date: Fri, 5 Feb 1999 13:06:13 +0200 (EET)
Subject: Re: Jforum: Boxing Columns
5-Feb-99 02:50 Paul Gauthier wrote:
> m42=:<"1@|: NB.Box each column of a matrix
>
Another way of understanding phrase "box columns" is
<@,."1@|: i.3 3
+-+-+-+
|0|1|2|
|3|4|5|
|6|7|8|
+-+-+-+
> "Nichols, Peter" wrote:
> > I know I could transpose but it seems like there should be a
> > more elegant way.
The only thing that is more elegent than Transpose is dyadic transpose.
nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 5 08:13:34 1999
From: "Chris Burke" <cdburke@interlog.com>
Subject: Re: Jforum: fmt
Date: Fri, 5 Feb 1999 07:56:03 -0500
charset="iso-8859-1"
Milan Ondrus writes:
> I'm looking at the utility fmt in the script <format>
>
> 'b4' fmt 3456 gives the char-matrix 3456
>
> but
>
> 'z4' fmt 3456 gives the char-matrix 0456 ... inserts a zero at the first
>position and wipes out the 3.
>
> 'z3' fmt 345 gives 045 ... again the first position disappears.
>
> This z-formatting seems to behave inconsistent with b-formatting or do I
>see that wrong?
This is a bug and a replacement script which includes Andrew Nikitin's
suggestions is available at www.jsoftware.com/source/format.zip
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 5 10:21:45 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: Boxing Columns
Date: Fri, 5 Feb 1999 10:06:24 -0500
->But how would you box the columns
>of a matrix? I know I could transpose but it seems like there should be a
>more elegant way.
I thought that if you extracted each column and boxed it, without doing the
transpose, you might use less space, since only one copy of the whole
array would be required. On a large array this would still be a loser,
since the
transpose could have better hit ratio in cache and virtual memory. But
when I tried it:
a =. (<@{~ <@(a:&;)"0@i.@{:@$)
I found that this version took more space than <"1@|: . Just a little
more: each one takes about twice the space of the original matrix.
At least in J4.01, which seems to accelerate <"1 .
Henry Rich
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 5 12:30:46 1999
Date: Fri, 05 Feb 1999 11:14:11 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Boxing Columns
References: <36BAB0EA.18772AE@sympatico.ca> <ABb2jksCL9@voxel.kharkov.ua>
Understanding is nice...Timing is also nice...
m42=:<"1@|:
m42x=:<@,."1@|:
Timing:
10000(6!:2)'m42 i.3 3'
0.000104
10000(6!:2)'m42x i.3 3'
0.000121
Andrew Nikitin wrote:
> 5-Feb-99 02:50 Paul Gauthier wrote:
> > m42=:<"1@|: NB.Box each column of a matrix
> >
>
> Another way of understanding phrase "box columns" is
>
> <@,."1@|: i.3 3
> +-+-+-+
> |0|1|2|
> |3|4|5|
> |6|7|8|
> +-+-+-+
>
> > "Nichols, Peter" wrote:
> > > I know I could transpose but it seems like there should be a
> > > more elegant way.
>
> The only thing that is more elegent than Transpose is dyadic transpose.
>
> nsg
>
> --------------------------------------------------------------------------------
> 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 Feb 6 09:25:04 1999
References: <36BB26E3.549D0A13@sympatico.ca>
Organization: Voxel
From: "Andrew Nikitin" <nsg@voxel.kharkov.ua>
Date: Sat, 6 Feb 1999 15:55:00 +0200 (EET)
Subject: Jforum: ,.
Dictionary says:
"The fit conjunction (,.!.f) provides fill specified by the items of f."
In what cases such fill is applied? When arguments agree in length, then no
need in fill arise. When they don't -- length error occurs. ie:
(i.2) ,. (i.3)
|length error
| (i.2) ,.(i.3)
(i.2) ,.!.9 (i.3)
|length error
| (i.2) ,.!.9(i.3)
Is there an example of j phrase containing ,.!.f with the result depending on
f?
Why there exist such inconsistent thing as "length error"? Why not to reshape
arguments to common length as for ,: always?
nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 10:43:05 1999
Date: Sat, 06 Feb 1999 10:29:52 -0500
From: Cliff Reiter <reiterc@lafvax.lafayette.edu>
Subject: Re: Jforum: ,. Fill atoms
Organization: Lafayette College
Content-type: text/plain; charset=us-ascii
References: <36BB26E3.549D0A13@sympatico.ca> <ABqc4lsWUB@voxel.kharkov.ua>
Andrew Nikitin wrote:
>
> Dictionary says:
>
> "The fit conjunction (,.!.f) provides fill specified by the items of f."
>
> In what cases such fill is applied? When arguments agree in length, then no
Perhaps Stitch isn't really what you want. Stitch is handy, but often
I find I really need adjoin when I thought about stitch first.
Consider the following examples
]c1=.,.1 2 3 NB. col 1
1
2
3
]c2=.,.10*1 2 3 4 5 NB. col 2
10
20
30
40
50
pca=. ,!. _ &.|: NB. padded column adjoin using infinity for fill
c1 pca c2
1 10
2 20
3 30
_ 40
_ 50
(i.3 3) pca i.4 4
0 1 2 0 1 2 3
3 4 5 4 5 6 7
6 7 8 8 9 10 11
_ _ _ 12 13 14 15
Does pca do what you want?
Also consider stitch with fill:
(i.2 3 4) ,.!. _] 10*i.2 3 7
0 1 2 3 _ _ _
4 5 6 7 _ _ _
8 9 10 11 _ _ _
0 10 20 30 40 50 60
70 80 90 100 110 120 130
140 150 160 170 180 190 200
12 13 14 15 _ _ _
16 17 18 19 _ _ _
20 21 22 23 _ _ _
210 220 230 240 250 260 270
280 290 300 310 320 330 340
350 360 370 380 390 400 410
> need in fill arise. When they don't -- length error occurs. ie:
>
> (i.2) ,. (i.3)
> |length error
> | (i.2) ,.(i.3)
> (i.2) ,.!.9 (i.3)
> |length error
> | (i.2) ,.!.9(i.3)
>
> Is there an example of j phrase containing ,.!.f with the result depending on
> f?
>
> Why there exist such inconsistent thing as "length error"? Why not to reshape
> arguments to common length as for ,: always?
>
> nsg
--
Clifford A. Reiter
Mathematics Department, Lafayette College
Easton, PA 18042 USA, 610-330-5277
^^^ Note new phone exchange
http://www.lafayette.edu/~reiterc
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 11:10:51 1999
Date: Sat, 6 Feb 1999 17:56:45 +0200 (EET)
From: Oleg Kobchenko <gccinc@usa.net>
Subject: Re: Jforum: Boxing Columns
In-Reply-To: <36BB26E3.549D0A13@sympatico.ca>
Timing is pay for what you wanna get. In Nikitin's
case we get added value of visually preserving the shape.
Consider:
<"1@|: i.3 3
+-----+-----+-----+
|0 3 6|1 4 7|2 5 8|
+-----+-----+-----+
<@,."1@|: i.3 3
+-+-+-+
|0|1|2|
|3|4|5|
|6|7|8|
+-+-+-+
NB. Here we add value afterwards
,.&.>@:<"1@|: i.3 3
+-+-+-+
|0|1|2|
|3|4|5|
|6|7|8|
+-+-+-+
NB. Hence the timing
10000(6!:2)'<"1@|: i.3 3'
0.0001342
10000(6!:2)',.&.>@:<"1@|: i.3 3'
0.0001913
10000(6!:2)'<@,."1@|: i.3 3'
0.0001713
On Fri, 5 Feb 1999, Paul Gauthier wrote:
> Understanding is nice...Timing is also nice...
>
> m42=:<"1@|:
> m42x=:<@,."1@|:
>
> Timing:
>
> 10000(6!:2)'m42 i.3 3'
> 0.000104
>
> 10000(6!:2)'m42x i.3 3'
> 0.000121
>
> Andrew Nikitin wrote:
>
> > 5-Feb-99 02:50 Paul Gauthier wrote:
> > > m42=:<"1@|: NB.Box each column of a matrix
> > >
> >
> > Another way of understanding phrase "box columns" is
> >
> > <@,."1@|: i.3 3
> > +-+-+-+
> > |0|1|2|
> > |3|4|5|
> > |6|7|8|
> > +-+-+-+
> >
> > > "Nichols, Peter" wrote:
> > > > I know I could transpose but it seems like there should be a
> > > > more elegant way.
> >
> > The only thing that is more elegent than Transpose is dyadic transpose.
> >
> > nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 12:10:42 1999
From: "Bjorn G. Helgason" <gosi@centrum.is>
Subject: RE: Jforum: Amending columns in a matrix
Date: Sat, 6 Feb 1999 16:56:48 -0000
Content-Type: multipart/mixed; boundary="---- =_NextPart_000_01BE51F1.F02B72A0"
------ =_NextPart_000_01BE51F1.F02B72A0
From: d.alis[SMTP:dalis@balcab.ch]
> a=: 1 3 5
> Anoun =. a
> Averb =. a"_
> 100 200 300 Anoun } i. 6 3
> 100 200 300 Averb } i. 6 3
Nice touch !!!
/Gosi
------ =_NextPart_000_01BE51F1.F02B72A0
Content-Type: application/ms-tnef
Content-Transfer-Encoding: base64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------ =_NextPart_000_01BE51F1.F02B72A0--
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 12:35:22 1999
Date: Sat, 06 Feb 1999 11:26:40 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: Boxing Columns
References: <Pine.LNX.4.04.9902061747430.2148-100000@spin.vl.net.ua>
Timing is pay for what I wanna get...
Well ! and what about that for a visual effect...
<"2,"0|:i.3 3
+-+-+-+
|0|1|2|
|3|4|5|
|6|7|8|
+-+-+-+
10000(6!:2)'<"2,"0|:i.3 3'
0.000159
10000(6!:2)'<@,."1@|: i.3 3'
0.00017
Oleg Kobchenko wrote:
> Timing is pay for what you wanna get. In Nikitin's
> case we get added value of visually preserving the shape.
> Consider:
>
> <"1@|: i.3 3
> +-----+-----+-----+
> |0 3 6|1 4 7|2 5 8|
> +-----+-----+-----+
> <@,."1@|: i.3 3
> +-+-+-+
> |0|1|2|
> |3|4|5|
> |6|7|8|
> +-+-+-+
>
> NB. Here we add value afterwards
>
> ,.&.>@:<"1@|: i.3 3
> +-+-+-+
> |0|1|2|
> |3|4|5|
> |6|7|8|
> +-+-+-+
>
> NB. Hence the timing
>
> 10000(6!:2)'<"1@|: i.3 3'
> 0.0001342
>
> 10000(6!:2)',.&.>@:<"1@|: i.3 3'
> 0.0001913
>
> 10000(6!:2)'<@,."1@|: i.3 3'
> 0.0001713
>
> On Fri, 5 Feb 1999, Paul Gauthier wrote:
>
> > Understanding is nice...Timing is also nice...
> >
> > m42=:<"1@|:
> > m42x=:<@,."1@|:
> >
> > Timing:
> >
> > 10000(6!:2)'m42 i.3 3'
> > 0.000104
> >
> > 10000(6!:2)'m42x i.3 3'
> > 0.000121
> >
> > Andrew Nikitin wrote:
> >
> > > 5-Feb-99 02:50 Paul Gauthier wrote:
> > > > m42=:<"1@|: NB.Box each column of a matrix
> > > >
> > >
> > > Another way of understanding phrase "box columns" is
> > >
> > > <@,."1@|: i.3 3
> > > +-+-+-+
> > > |0|1|2|
> > > |3|4|5|
> > > |6|7|8|
> > > +-+-+-+
> > >
> > > > "Nichols, Peter" wrote:
> > > > > I know I could transpose but it seems like there should be a
> > > > > more elegant way.
> > >
> > > The only thing that is more elegent than Transpose is dyadic transpose.
> > >
> > > nsg
>
> --------------------------------------------------------------------------------
> 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 Feb 6 15:15:08 1999
Date: Sat, 06 Feb 1999 14:07:49 -0600
From: Peter Nichols <pnichols@airmail.net>
Subject: Re: Jforum: Boxing Columns
References: <Pine.LNX.4.04.9902061747430.2148-100000@spin.vl.net.ua> <36BC7B4F.AE39B789@sympatico.ca>
Working on Saturday? Or just amusing yourself. Well, this idiom is the
strangest to my mind so far. I kind of like it though. So what are you
working at these days?
Oh, this is my personal email, the other is at work.
Ciao
Paul Gauthier wrote:
>
> Timing is pay for what I wanna get...
> Well ! and what about that for a visual effect...
> <"2,"0|:i.3 3
> +-+-+-+
> |0|1|2|
> |3|4|5|
> |6|7|8|
> +-+-+-+
> 10000(6!:2)'<"2,"0|:i.3 3'
> 0.000159
> 10000(6!:2)'<@,."1@|: i.3 3'
> 0.00017
>
> Oleg Kobchenko wrote:
>
> > Timing is pay for what you wanna get. In Nikitin's
> > case we get added value of visually preserving the shape.
> > Consider:
> >
> > <"1@|: i.3 3
> > +-----+-----+-----+
> > |0 3 6|1 4 7|2 5 8|
> > +-----+-----+-----+
> > <@,."1@|: i.3 3
> > +-+-+-+
> > |0|1|2|
> > |3|4|5|
> > |6|7|8|
> > +-+-+-+
> >
> > NB. Here we add value afterwards
> >
> > ,.&.>@:<"1@|: i.3 3
> > +-+-+-+
> > |0|1|2|
> > |3|4|5|
> > |6|7|8|
> > +-+-+-+
> >
> > NB. Hence the timing
> >
> > 10000(6!:2)'<"1@|: i.3 3'
> > 0.0001342
> >
> > 10000(6!:2)',.&.>@:<"1@|: i.3 3'
> > 0.0001913
> >
> > 10000(6!:2)'<@,."1@|: i.3 3'
> > 0.0001713
> >
> > On Fri, 5 Feb 1999, Paul Gauthier wrote:
> >
> > > Understanding is nice...Timing is also nice...
> > >
> > > m42=:<"1@|:
> > > m42x=:<@,."1@|:
> > >
> > > Timing:
> > >
> > > 10000(6!:2)'m42 i.3 3'
> > > 0.000104
> > >
> > > 10000(6!:2)'m42x i.3 3'
> > > 0.000121
> > >
> > > Andrew Nikitin wrote:
> > >
> > > > 5-Feb-99 02:50 Paul Gauthier wrote:
> > > > > m42=:<"1@|: NB.Box each column of a matrix
> > > > >
> > > >
> > > > Another way of understanding phrase "box columns" is
> > > >
> > > > <@,."1@|: i.3 3
> > > > +-+-+-+
> > > > |0|1|2|
> > > > |3|4|5|
> > > > |6|7|8|
> > > > +-+-+-+
> > > >
> > > > > "Nichols, Peter" wrote:
> > > > > > I know I could transpose but it seems like there should be a
> > > > > > more elegant way.
> > > >
> > > > The only thing that is more elegent than Transpose is dyadic transpose.
> > > >
> > > > nsg
> >
> > --------------------------------------------------------------------------------
> > 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
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 17:24:22 1999
From: "Donald Pittenger" <dbpitt@demlab.com>
Subject: Jforum: 2-D graph modes
Date: Sat, 6 Feb 1999 14:13:25 -0800
boundary="----=_NextPart_000_0008_01BE51DA.DC1B2740"
This is a multi-part message in MIME format.
------=_NextPart_000_0008_01BE51DA.DC1B2740
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
I am doing some gl2 graphing, and drafted a program using J402a under =
Win98.
At another location, I ran the same program using J402a, but this time =
under NT4.0.
The latter graph had bold lines that should have been uniform width (as =
they were in the Win98 setting) stretched in the vertical direction...as =
if they were built from tiny character blocs rather than square pixels.
So far as I know, the J environments are identical, aside from the =
operating system and the fact that the case 1 computer is a Compaq =
300Mhz Pentium and case 2 is a Dell 400Mhz Pentium. Also, the case 2 =
computer's J was downloaded a few days before that for case 1, if that =
means anything.
Does anyone have any suggestions as to how to get those lines back to =
constant width? The wd on-line docs suggest the default glmap ought to =
deal with this without intervention.
Thanks for your help (in advance).
Don Pittenger
------=_NextPart_000_0008_01BE51DA.DC1B2740
charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE HTML PUBLIC "-//W3C//DTD W3 HTML//EN">
<HTML>
<HEAD>
<META content=3Dtext/html;charset=3Diso-8859-1 =
http-equiv=3DContent-Type>
<META content=3D'"MSHTML 4.72.3110.7"' name=3DGENERATOR>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT color=3D#000000 size=3D2>I am doing some gl2 graphing, and =
drafted a=20
program using J402a under Win98.</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>At another location, I ran the same =
program=20
using J402a, but this time under NT4.0.</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>The latter graph had bold lines that =
should have=20
been uniform width (as they were in the Win98 setting) stretched in the =
vertical=20
direction...as if they were built from tiny character blocs rather than =
square=20
pixels.</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>So far as I know, the J environments =
are=20
identical, aside from the operating system and the fact that the =
case 1=20
computer is a Compaq 300Mhz Pentium and case 2 is a Dell 400Mhz =
Pentium. =20
Also, the case 2 computer's J was downloaded a few days before that for =
case 1,=20
if that means anything.</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>Does anyone have any suggestions as =
to how to=20
get those lines back to constant width? The wd on-line docs =
suggest the=20
default glmap ought to deal with this without intervention.</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>Thanks for your help (in =
advance).</FONT></DIV>
<DIV><FONT color=3D#000000 size=3D2></FONT> </DIV>
<DIV><FONT color=3D#000000 size=3D2>Don =
Pittenger</FONT></DIV></BODY></HTML>
------=_NextPart_000_0008_01BE51DA.DC1B2740--
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sat Feb 6 22:51:04 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: ,.
Date: Sat, 6 Feb 1999 22:39:48 -0500
>Is there an example of j phrase containing ,.!.f with the result depending
on
>f?
(i. 3 3 2) ,.!._ i. 3 1
0 1
2 3
4 5
0 _
6 7
8 9
10 11
1 _
12 13
14 15
16 17
2 _
Henry Rich
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Sun Feb 7 04:14:54 1999
Date: Sun, 07 Feb 1999 03:05:09 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: ,. Fill atoms
References: <36BB26E3.549D0A13@sympatico.ca> <ABqc4lsWUB@voxel.kharkov.ua> <36BC5FF0.6234@lafvax.lafayette.edu>
I agree with Cliff and will just add a little touch.
v2pca=:|:@:(,:!.9) NB. vectors to padded column adjoin using 9 for fill
(i.2)v2pca(i.3)
0 0
1 1
9 2
Why ?, my old habit of timing again...
10000(6!:2)'(,"0 i.2)pca(i.3)'
0.000203
10000(6!:2)'(i.2)v2pa(i.3)'
0.000138
Cliff Reiter wrote:
> Andrew Nikitin wrote:
> >
> > Dictionary says:
> >
> > "The fit conjunction (,.!.f) provides fill specified by the items of f."
> >
> > In what cases such fill is applied? When arguments agree in length, then no
>
> Perhaps Stitch isn't really what you want. Stitch is handy, but often
> I find I really need adjoin when I thought about stitch first.
>
> Consider the following examples
>
> ]c1=.,.1 2 3 NB. col 1
> 1
> 2
> 3
> ]c2=.,.10*1 2 3 4 5 NB. col 2
> 10
> 20
> 30
> 40
> 50
> pca=. ,!. _ &.|: NB. padded column adjoin using infinity for fill
> c1 pca c2
> 1 10
> 2 20
> 3 30
> _ 40
> _ 50
>
> (i.3 3) pca i.4 4
> 0 1 2 0 1 2 3
> 3 4 5 4 5 6 7
> 6 7 8 8 9 10 11
> _ _ _ 12 13 14 15
>
> Does pca do what you want?
> Also consider stitch with fill:
>
> (i.2 3 4) ,.!. _] 10*i.2 3 7
> 0 1 2 3 _ _ _
> 4 5 6 7 _ _ _
> 8 9 10 11 _ _ _
> 0 10 20 30 40 50 60
> 70 80 90 100 110 120 130
> 140 150 160 170 180 190 200
>
> 12 13 14 15 _ _ _
> 16 17 18 19 _ _ _
> 20 21 22 23 _ _ _
> 210 220 230 240 250 260 270
> 280 290 300 310 320 330 340
> 350 360 370 380 390 400 410
>
> > need in fill arise. When they don't -- length error occurs. ie:
> >
> > (i.2) ,. (i.3)
> > |length error
> > | (i.2) ,.(i.3)
> > (i.2) ,.!.9 (i.3)
> > |length error
> > | (i.2) ,.!.9(i.3)
> >
> > Is there an example of j phrase containing ,.!.f with the result depending on
> > f?
> >
> > Why there exist such inconsistent thing as "length error"? Why not to reshape
> > arguments to common length as for ,: always?
> >
> > nsg
> --
> Clifford A. Reiter
> Mathematics Department, Lafayette College
> Easton, PA 18042 USA, 610-330-5277
> ^^^ Note new phone exchange
> http://www.lafayette.edu/~reiterc
>
> --------------------------------------------------------------------------------
> 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 Mon Feb 8 16:44:18 1999
From: "Chris Burke" <cdburke@interlog.com>
Subject: Jforum: J402 for Mac
Date: Mon, 8 Feb 1999 16:24:12 -0500
charset="iso-8859-1"
J402 for the Mac (PowerPC and 68K) is now available at www.jsoftware.com
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 9 11:06:27 1999
References: <36BC5FF0.6234@lafvax.lafayette.edu>
Organization: Voxel
From: "Andrew Nikitin" <nsg@voxel.kharkov.ua>
Date: Tue, 9 Feb 1999 16:13:03 +0200 (EET)
Subject: Re: Jforum: ,. Fill atoms
6-Feb-99 10:29 Cliff Reiter wrote:
> Also consider stitch with fill:
>
> (i.2 3 4) ,.!. _] 10*i.2 3 7
> 0 1 2 3 _ _ _
> 4 5 6 7 _ _ _
> 8 9 10 11 _ _ _
> 0 10 20 30 40 50 60
> 70 80 90 100 110 120 130
> 140 150 160 170 180 190 200
>
> 12 13 14 15 _ _ _
> 16 17 18 19 _ _ _
> 20 21 22 23 _ _ _
> 210 220 230 240 250 260 270
> 280 290 300 310 320 330 340
> 350 360 370 380 390 400 410
I have catch it up: catenate (and stitch) applies fill element to pad items to
common shape, but it doesn't ravel frames (_1 frames in the case of ,.), so
length error occurs when frames disagree in length. And this error occurs not
"in" , (or ,.) but in " (,.=:,"_1 , as dictionary claims)
This job is rather for " conjunction, who executes implicit
looping over left and right operands simultaneously. So, it is " or something
in place of it, who must create additional fill items for the shorter
argument.
nsg
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Tue Feb 9 12:20:49 1999
Date: Tue, 09 Feb 1999 10:58:55 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: ,. Fill atoms
References: <36BC5FF0.6234@lafvax.lafayette.edu> <ABl94msycH@voxel.kharkov.ua>
Remember my prior message ?
I agree with Cliff and will just add a little touch.
v2pca=:|:@:(,:!.9) NB. vectors to padded column adjoin using 9 for fill
(i.2)v2pca(i.3)
0 0
1 1
9 2
Why ?, my old habit of timing again...
10000(6!:2)'(,"0 i.2)pca(i.3)'
0.000203
10000(6!:2)'(i.2)v2pa(i.3)'
0.000138
Well, you could have two verbs, let say:
v0=:(,:!.9)
v1=:|:@:v0
An atomic argument in x,:y is first reshaped to the shape of the other (or to a list
if the other argument is also atomic); the results are then itemized and catenated,
as in (,:x),(,:y) .The fit conjunction (,:!.f) provides fill specified by the items
of f.
As per the vocabulary. I hope this fits your need.../Paul
Andrew Nikitin wrote:
> 6-Feb-99 10:29 Cliff Reiter wrote:
> > Also consider stitch with fill:
> >
> > (i.2 3 4) ,.!. _] 10*i.2 3 7
> > 0 1 2 3 _ _ _
> > 4 5 6 7 _ _ _
> > 8 9 10 11 _ _ _
> > 0 10 20 30 40 50 60
> > 70 80 90 100 110 120 130
> > 140 150 160 170 180 190 200
> >
> > 12 13 14 15 _ _ _
> > 16 17 18 19 _ _ _
> > 20 21 22 23 _ _ _
> > 210 220 230 240 250 260 270
> > 280 290 300 310 320 330 340
> > 350 360 370 380 390 400 410
>
> I have catch it up: catenate (and stitch) applies fill element to pad items to
> common shape, but it doesn't ravel frames (_1 frames in the case of ,.), so
> length error occurs when frames disagree in length. And this error occurs not
> "in" , (or ,.) but in " (,.=:,"_1 , as dictionary claims)
>
> This job is rather for " conjunction, who executes implicit
> looping over left and right operands simultaneously. So, it is " or something
> in place of it, who must create additional fill items for the shorter
> argument.
>
> nsg
>
> --------------------------------------------------------------------------------
> 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 Tue Feb 9 16:49:30 1999
Date: Tue, 09 Feb 1999 15:32:33 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Re: Jforum: ,. Fill atoms
References: <36BC5FF0.6234@lafvax.lafayette.edu> <ABl94msycH@voxel.kharkov.ua>
Please help me understand what exactly you are trying to achieve...
For example:
m=:i.2 3
n=.i.4
l=:,!._ NB. append lines
c=:|:@:((|:@[),!._|:@]) NB. append columns
m l n
0 1 2 _
3 4 5 _
0 1 2 3
m c n
0 1 2 0
3 4 5 1
_ _ _ 2
_ _ _ 3
is it something like the above ?...Regards/Paul
Andrew Nikitin wrote:
> 6-Feb-99 10:29 Cliff Reiter wrote:
> > Also consider stitch with fill:
> >
> > (i.2 3 4) ,.!. _] 10*i.2 3 7
> > 0 1 2 3 _ _ _
> > 4 5 6 7 _ _ _
> > 8 9 10 11 _ _ _
> > 0 10 20 30 40 50 60
> > 70 80 90 100 110 120 130
> > 140 150 160 170 180 190 200
> >
> > 12 13 14 15 _ _ _
> > 16 17 18 19 _ _ _
> > 20 21 22 23 _ _ _
> > 210 220 230 240 250 260 270
> > 280 290 300 310 320 330 340
> > 350 360 370 380 390 400 410
>
> I have catch it up: catenate (and stitch) applies fill element to pad items to
> common shape, but it doesn't ravel frames (_1 frames in the case of ,.), so
> length error occurs when frames disagree in length. And this error occurs not
> "in" , (or ,.) but in " (,.=:,"_1 , as dictionary claims)
>
> This job is rather for " conjunction, who executes implicit
> looping over left and right operands simultaneously. So, it is " or something
> in place of it, who must create additional fill items for the shorter
> argument.
>
> nsg
>
> --------------------------------------------------------------------------------
> 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 Wed Feb 10 11:14:23 1999
Date: Wed, 10 Feb 1999 09:52:45 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Jforum: Shame on me...
For those APL old timers, enclose on axis (origin 0 of course) could be
simulated in J with this:
ea=:4 :'<"($,x.)]:x.|:y.'
But this should be used for getting used to J, the timing on this one is
terrible...Regards/Paul
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 06:32:27 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Jforum: How do I find Mapi.dll?
Date: Fri, 12 Feb 1999 06:13:26 -0500
I am trying to send Email from a J program.
I decided that the MAPI interface was the one to use (comments
on that decision welcomed), so I wrote the program, ending with
mailrc =. 'Mapi MAPISendMail i i i *m i i' cd 0;0;msgallo;0;0
but this returns _1, indicating 'library not found'. I see the library in
c:\Windows\System\Mapi.dll. What do I need to do to access
the library? I have tried specifying 'Mapi.dll' and
'C:\Windows\System\Mapi' and 'C:\Windows\System\Mapi.dll'
in the argument to cd, without success.
Henry Rich
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 09:02:49 1999
Date: Fri, 12 Feb 1999 07:46:59 -0600
From: Paul Gauthier <pap.gauthier@sympatico.ca>
Subject: Jforum: Oups! errata...
In my prior message, winsck.dll should have been written winsck.ocx (the
utility registers both .dll and .ocx).
Regards/Paul
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 09:11:58 1999
Date: Fri, 12 Feb 1999 13:48:36 GMT
From: J E H Shaw <strgh@csv.warwick.ac.uk>
Subject: Jforum: WGL - J scripts for graphics output (preliminary versions!)
Dear All,
On my homepage are some tentative J scripts implementing WGL,
a simple graphics language with ISIG & postscript output.
I find the scripts useful for including graphics within LaTeX
documents (you'll need psfrag, available from the ctan archives,
if you have graphics containing text that needs interpreting by TeX).
The only current documentation is a 5-year-old WGL manual describing
the language + an implementation in APL*PLUS & tpic,
but all my J code is of course self-documenting anyway :-)
See http://www.warwick.ac.uk/statsdept/Staff/JEHS/jehwgl.htm
Regards, Ewart Shaw
PS: The above is all using J3. After a rough year I hope to be
able to get & use J4 soon, and work again on my 'Warwick
Guide to J'. Thanks again to everyone who commented on it,
and please forgive my rudeness if I forgot to respond to you
personally.
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 09:13:16 1999
Date: Fri, 12 Feb 1999 13:51:48 GMT
From: J E H Shaw <strgh@csv.warwick.ac.uk>
Subject: Jforum: Problems with =. =: and timex
I've encountered awkward problems while writing a verb that
automatically obtains timings for each line of a second
(defined) verb. The following is the best I've come up with
(invoked by timefn 'fun';y or timefn x;'fun';y):
timefn=: verb define
50 timefn y.
:
xTMP=: 0 pick y.
fTMP=. ;/ ('';1;1;1) {:: 5!:1 ] _2{y.
yTMP=: _1 pick y.
(x.&timeline each ,. ]) fTMP
)
timeline=: dyad define
sTMP=. (y. rplc 'y.';'yTMP') rplc 'x.';'xTMP'
aTMP=. x. timex sTMP rplc '=:';'=.'
do sTMP rplc '=.';'=:'
aTMP
)
I still seem to need global assignments (=:)
even after replacing the `timeline each' in timefn by a loop
to keep all invocations of timex within a single verb:
NB. replace '(x.&timeline each ,. ]) fTMP' by...
tTMP=. i.0
for_iTMP. i. # fTMP do.
sTMP=. ((iTMP{fTMP) rplc 'y.';'yTMP') rplc 'x.';'xTMP'
tTMP=. tTMP, x. timex sTMP rplc '=:';'=.'
do sTMP rplc '=.';'=:'
end.
(;/tTMP) ,. ;/fTMP
)
Can anyone suggest how to write a timefn verb without
creating/overwriting globals? It must be able to handle
lines like 'x.=. x. , 0' in the 'fun'-:fTMP function.
A related query concerns the inability to create local variables
in a specified locale (to tidy up the looping version of timefn):
NB. attempt to create local y_tmp_
(3 : 'y_tmp_=: y.') 1
1
(3 : 'y_tmp_=. y.') 2
|nonce error
| y_tmp_ =.y.
y_tmp_
1
Similarly I can't have 'for_i_tmp_. ... do. ... end.'
I imagine that this sort of locale abuse might be frowned upon,
but is the above behaviour explained in the J documentation?
(I may well have overlooked or misunderstood it!)
-- Ewart Shaw
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 09:34:57 1999
From: M.Day@fscii.maff.gov.uk
Fri, 12 Feb 1999 13:54:50 +0000
Fri, 12 Feb 1999 14:15:37 +0000
Fri, 12 Feb 1999 14:15:04 +0000
Date: Fri, 12 Feb 1999 14:15:04 +0000
Content-Identifier: m1210212141447aa
Alternate-Recipient: Allowed
In-Reply-To: <199902121348.NAA08272@mimosa.csv.warwick.ac.uk>
Subject: Re: Jforum: WGL - J scripts for graphics output (preliminary versions!)
Ewart
(sorry to broadcast this, J-listers, but I've lost
Ewart's e-address.)
Reminds me of BAPLA's APL Stats Library (ASL) etc -
is anybody maintaining it, adding to it ...?
I remember WGL (or similar) being extolled by the
Warwickmen.
Mike
215pm
12 2 99
(m.day@fscii.maff.gov.uk)
JEHS said:
>Dear All,
>On my homepage are some tentative J scripts implementing WGL,
>.....
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 09:47:46 1999
From: "Seymour Glass" <glasss@mindspring.com>
Subject: Re: Jforum: How do I find Mapi.dll? - found
Date: Fri, 12 Feb 1999 09:21:43 -0500
Oh, I see now, it's Mapi32.dll
Henry Rich
-----Original Message-----
From: Seymour Glass <glasss@mindspring.com>
Date: Friday, February 12, 1999 06:15
Subject: Jforum: How do I find Mapi.dll?
>I am trying to send Email from a J program.
>
>I decided that the MAPI interface was the one to use (comments
>on that decision welcomed), so I wrote the program, ending with
>
>mailrc =. 'Mapi MAPISendMail i i i *m i i' cd 0;0;msgallo;0;0
>
>but this returns _1, indicating 'library not found'. I see the library in
>c:\Windows\System\Mapi.dll. What do I need to do to access
>the library? I have tried specifying 'Mapi.dll' and
>'C:\Windows\System\Mapi' and 'C:\Windows\System\Mapi.dll'
>in the argument to cd, without success.
>
>Henry Rich
>
>
>
>---------------------------------------------------------------------------
-----
>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 Fri Feb 12 10:22:29 1999
Date: Fri, 12 Feb 1999 15:00:17 +0000 (GMT)
From: J E H Shaw <strgh@csv.warwick.ac.uk>
Subject: Re: Jforum: WGL - J scripts for graphics output
In-Reply-To: <m1210212141447aa*/I=M/S=Day/OU=FSCII/O=MAFF/PRMD=MAFF400/ADMD=ATTmail/C=GB/@MHS>
[Following up Mike Day's post about WGL & ASL (APL Stats Language]
WGL started life as `gr', an unholy mixture of C,
Fortran 77 & 68000 assembler I wrote in 1985.
Many of the papers in `the Statistician' 36:2-3
(`Practical Bayesian Statistics') include gr output.
When I came to Warwick I translated gr to APL,
which then formed the basis for ASL.
I don't know if anyone still maintains or even uses ASL,
but I intend to add further J scripts with higher-level graphics,
using the basic WGL language, to my WWW pages. Real Soon Now.
-- Ewart Shaw
J.E.H.Shaw [Ewart Shaw] strgh@uk.ac.warwick TEL: +44 203 523069
Department of Statistics, University of Warwick, Coventry CV4 7AL, U.K.
http://www.warwick.ac.uk/statsdept/Staff/JEHS/
yacc - the piece of code that understandeth all parsing
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 10:39:50 1999
Date: Fri, 12 Feb 1999 15:03:24 +0000 (GMT)
From: J E H Shaw <strgh@csv.warwick.ac.uk>
cc: Receipt Notification Requested <forum@jsoftware.com>
Subject: Re: Jforum: WGL - J scripts for graphics output (preliminary versions!)
In-Reply-To: <m1210212141447aa*/I=M/S=Day/OU=FSCII/O=MAFF/PRMD=MAFF400/ADMD=ATTmail/C=GB/@MHS>
Dear Mike,
Thank you for your comment - I've sent a note about WGL, ASL etc.
to jforum. I'd be interested to hear your comments on WGL if you
should use it or just look at it - I'm not aware of any bugs,
but am painfully aware of meny inefficiencies.
Regards, Ewart
J.E.H.Shaw [Ewart Shaw] strgh@uk.ac.warwick TEL: +44 203 523069
Department of Statistics, University of Warwick, Coventry CV4 7AL, U.K.
http://www.warwick.ac.uk/statsdept/Staff/JEHS/
yacc - the piece of code that understandeth all parsing
--------------------------------------------------------------------------------
J Forum: for information about this list, see http://www.jsoftware.com/forum.htm
From owner-jsoftware@lists.interlog.com Fri Feb 12 14:39:33 1999
From: "John D. Baker" <bakerjd@kos.net>
Subject: Re: Jforum: How do I find Mapi.dll?
Date: Fri, 12 Feb 1999 14:20:46 -0500
charset="iso-8859-1"
I'm curious about the *m declaration in
your call. I've been doing a lot of J/WinAPI programming lately
and as far as I know the you can only declare:
c character (1 byte)
s short integer (2 byte)
i integer (4 byte)
f short floating-point (4 byte)
d floating point (8 byte)
j complex (16 byte) (not as result)