c =. 1 { a is problematic because if a is deleted, you are left holding
the entire a just to make one item available. That would waste space.
b =. , a could perhaps increase the usecount of the backer, but I didn't
implement it that way, and I think I decided it wouldn't work to, but I
can't remember why. Once you know about virtual blocks you might stop
writing b =. , a and use (,a) instead.
Henry Rich
On 5/16/2020 6:43 PM, bill lam wrote:
Shouldn't assignment like
b=. , a
c=. 1{a
just increase reference count of the mother instead of deep copy?
On Sun, May 17, 2020, 6:27 AM Henry Rich <[email protected]> wrote:
I wouldn't say so. Reshape takes virtually no time/space, but boxing or
assigning the reshaped result does. If you do something like
+/ @: (,/) y
the result of ,/ is never realized and it would be wrong to charge it
with time/space not used.
Similarly
(}. - }:) y
the }. and }: create virtual results that are never realized.
You need to expand your mental model beyond time/space for a verb, to
include time/space for realization when that becomes necessary.
Henry Rich
On 5/16/2020 1:22 PM, 'Mike Day' via Programming wrote:
Oh... so one needs to, say, assign the result to see the real t/s?
Thanks,
Mike
Sent from my iPad
On 16 May 2020, at 17:05, Henry Rich <[email protected]> wrote:
The time/space numbers are telling you that (x $ y) produces a virtual
result when it can, while (_2 ]\ y) doesn't (yet). If you use the result
immediately, the space saving is real. If you box the result or save it in
a name, the value will be realized and the space saving will vanish.
Henry Rich
On 5/16/2020 11:14 AM, 'Michael Day' via Programming wrote:
Much neater than what I was about to offer, unless Raoul needs to
specify the fill,
in which case, this alternative rather minimal amendment is worth
consideration:
($!._ ~2,~>.@-:@#) i.7
0 1
2 3
4 5
6 _
cf
_2]\ i.7
0 1
2 3
4 5
6 0
Also, the time and space performance _might_ be important for large
inputs:
ts' $ _2]\ list ' [list =: i.100000
0.000618 1.04986e6
ts'($!._ ~2,~>.@-:@#) list'
2.7e_6 2304
Cheers,
Mike
On 16/05/2020 15:57, 'Rob Hodgkinson' via Programming wrote:
You could try Infix … here with NuVoc link…
https://code.jsoftware.com/wiki/Vocabulary/bslash#dyadic
x u\ y where x is eg _2 means apply very b to successive pairs
(_ for non-overlapping).
_2 ]\ 1 2 3 4 5 6
1 2
3 4
5 6
_2 ]\ 1 2 3 4 5 6 7
1 2
3 4
5 6
7 0
_2 <\ 1 2 3 4 5 6
┌───┬───┬───┐
│1 2│3 4│5 6│
└───┴───┴───┘
HTH…/Rob
On 17 May 2020, at 12:42 am, Raoul Schorer <[email protected]>
wrote:
Hello,
I am convinced that this must be trivial, but I wasn't able to find
in the documentation how to reshape a list to a table without manually
extracting the length.
in summary, is there a more direct way of doing:
lst =. i. 6
((2,~2%~#) $ ]) lst
for a list of arbitrary length?
Thanks!
Raoul
----------------------------------------------------------------------
For information about J forums see
http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see
http://www.jsoftware.com/forums.htm
--
This email has been checked for viruses by AVG.
https://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
--
This email has been checked for viruses by AVG.
https://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
