IMO the current behavior is acceptable for
if. 0 > _. do. etc…
The user who wrote this sentence must have some special task to accomplish
and can just skip the condition depending on what he/she think the
condition should be true or false or signaling error.
If the _. is a result of operation on
I don't know how often, but there is an application of higher-ranked
arrays: FFT. http://code.jsoftware.com/wiki/Essays/FFT . For argument
vectors of length 2^n, the algorithm creates rank n binary hypercubes,
shape n$2. (Subfunction "cube".)
On Fri, Dec 29, 2017 at 8:10 PM, TongKe Xue wrote
I now remember why I hadn’t used j806; my Mac is too old for AVX.
Anyhow, I tried j806 without AVX:
2 7 p.1e308j_
_j_
So, apparently the problem has already been solved. Sorry for the bother!
Louis
PS: I still would make something throw an error when doing things like
if. 0 > _. do. etc…
for
Hi,
1. I agree that the concept of rank + cell + frames + implicit
parallel8ism is very cool.
2. 5 dimension is commonly defined as N, C, D, H, W:
N = number of images
C = channels
D = depth
H = height
W = width
In practice, how often do we use algorithms with > 5 dim tensors?
J's position on _. is that it is an abomination and we make no claims
for what happens when _. appears in an argument and spend no time
worrying about it.
Henry Rich
On 12/29/2017 10:50 PM, Louis de Forcrand wrote:
Sorry for not trying this in j806, I should install it on my Mac.
Will update.
Sorry for not trying this in j806, I should install it on my Mac.
Will update.
As for control structures treating _. as false, I was mistaken.
I was actually comparing a float with _., and that returned 0.
Nevertheless, I think control structures should signal _. instead of silently
accepting it,
This is what I got from J806
foo
3 : 0
if. _ - _ do.
echo 1
else.
echo 0
end.
EMPTY
)
foo''
|NaN error: foo
| _-_
bar
3 : 0
if. _. do.
echo 1
else.
echo 0
end.
EMPTY
)
bar''
1
foo is reasonable.
bar is also reasonable to me, because _. is not equal 0 so that
the condition is
What about control structures treating _. as false?
I personally would make that an error, and I would also say that code that
makes use of this should be considered bad style.
Louis
> On 29 Dec 2017, at 18:29, Henry Rich wrote:
>
> Surely not intended. Put it on the bug list.
>
> Henry Rich
Surely not intended. Put it on the bug list.
Henry Rich
On 12/29/2017 6:06 PM, Louis de Forcrand wrote:
On the Indeterminate vocabulary page of the dictionary it is written
The only ways to create _. are as follows:
direct entry of _.
".
3!:n
DLL call (database
On the Indeterminate vocabulary page of the dictionary it is written
The only ways to create _. are as follows:
direct entry of _.
".
3!:n
DLL call (database, LAPACK, etc.)
and
Primitives on arguments not containing _. signal NaN error instead of producing
_. .
$T
49
5 10$ a. i. T
84 104 101 32 99 97 116 39 115 32
99 111 108 108 97 114 32 105 115 32
108 97 121 105 110 103 32 111 117 116
115 105 100 101 32 66 105 108 108 39
115 32 112 111 114 99 104 46 10 84
5 10$ a. i. U
84 104 101 32 99 97 116 39 115 32
99 111 108 108
T has a line feed at the end.
T-:U,LF
1
On Fri, Dec 29, 2017 at 8:47 AM, Linda Alvord
wrote:
> What is in T, that isn't in U?
>
> T=: 0 :0
> The cat's collar is laying outside Bill's porch.
> )
>
>
>U=:'The cat''s collar is laying outside Bill''s porch.'
>T-:U
> 0
>,. T;U
> ┌───
What is in T, that isn't in U?
T=: 0 :0
The cat's collar is laying outside Bill's porch.
)
U=:'The cat''s collar is laying outside Bill''s porch.'
T-:U
0
,. T;U
┌─┐
│The cat's collar is laying outside Bill's porch. │
├───
13 matches
Mail list logo
