Dan - it _was_ an accurate transcription. However, I neglected to check the
definition of "each":
each
"0
in my session. If I substitute what I assumed it was, &.>, the whole
problem makes a lot more sense.
Using the trick you taught me, we find the offending redefiner:
(4!:4{.;:'each') {:: 4!:3''
c:\program files\j601\system\extras\util\browser.ijs
On 3/30/07, Dan Bron <[EMAIL PROTECTED]> wrote:
Devon,
Are you sure the message quoted below contains an accurate transcript of
your J session? Here's what I get:
table =. ,. each <"1 & |: i. 4 3
(>,.&.>/table) ; |:;"2 table
+-------+-------------------------+
|0 1 2|0 3 6 9 1 4 7 10 2 5 8 11|
|3 4 5| |
|6 7 8| |
|9 10 11| |
+-------+-------------------------+
(>,.&.>/table) -: |:;"2 table
0
$|:;"2 table
1 12
$>,.&.>/table
4 3
$table
3
I think this last result is the crux: your table and my table are
different. I have no unit axes. Mine is a simple vector of 3 boxes, each
containing a single column of numbers (each with shape 4 1 ).
A formulation equivalent to yours, but with the noun table as given,
would be:
|:@:(,"2)>table
0 1 2
3 4 5
6 7 8
9 10 11
or:
,"2@:(0&|:)@:> table
0 1 2
3 4 5
6 7 8
9 10 11
in either case, you could use {. (head) in place of , (ravel) .
And you're right, as far as I can deduce, these formulations give results
equivalent to >@:(,&.>)/ (with exceptions for corner cases like 0 e. $
table ?) . And the performance metrics may differ, because there are
many, small intermediate arrays in the ,. solution, as opposed to the few,
large ones in the |: solutions.
-Dan
----- Original Message ---------------
Subject: Re: [Jprogramming] Unboxing Columns
From: "Devon McCormick" <[EMAIL PROTECTED]>
Date: Fri, 30 Mar 2007 15:23:31 -0400
To: "Programming forum" <[email protected]>
>Well, admittedly,
>
> table =. ,. each <"1 & |: i. 4 3
> (>,.&.>/table) ; |:;"2 table NB. The results look the same
>+-------+-------+
>|0 1 2|0 1 2|
>|3 4 5|3 4 5|
>|6 7 8|6 7 8|
>|9 10 11|9 10 11|
>+-------+-------+
> (>,.&.>/table) -: |:;"2 table NB. but aren't
>0
> $|:;"2 table NB. because
>4 3
> $>,.&.>/table
>1 1 4 3
>
>In my experience, extra, singleton axes are as much a
>nuisance as anything else.
>
>Of course, maybe they are present in the original array for a reason
>and need to be retained?
>
> $table
>3 1 1
>
>
>On 3/30/07, Dan Bron <[EMAIL PROTECTED]> wrote:
>>
>> RM = Raul Miller, DM = Devon McCormick
>>
>> RM> Personally, I would phrase that as
>> RM> >,.&.>/table
>>
>> DM>I think
>> DM> |:;"2 table
>>
>> Tsk. That's like recommending a book you haven't read. Try running
that
>> line...
>>
>> By the way, Leigh, if you promise that table is a vector of at least
two
>> boxes, then you can shorten your verb a bit:
>>
>> ,.&:>/ table
>>
>> But the general case is much nicer with >@:(,.&.>/) which is faster
and
>> leaner anyway.
>>
>> -Dan
>> ----------------------------------------------------------------------
>> For information about J forums see http://www.jsoftware.com/forums.htm
>>
>
>
>
>--
>Devon McCormick, CFA
>^me^ at acm.
>org is my
>preferred e-mail
>----------------------------------------------------------------------
>For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
--
Devon McCormick, CFA
^me^ at acm.
org is my
preferred e-mail
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
