Hi Louis,
so I suppose this was not an issue.
BTW if you believe a lambda behaves differently than a defined function then
you can check the lambda with *⎕CR*, e.g.:
*ADD←{⍺+⍵;C;⎕IO;D}**
** ⎕CR 'ADD'**
**λ←⍺ λ1 ⍵;C;⎕IO;D*
*λ←⍺+⍵ *
Or, in your example:
* f←{⍎(⍵>1)⊃'⍵' '⍵×⍙ ⍵-1'⊣⎕IO←0;⎕IO}**
** ⎕CR 'f'**
**λ←λ1 ⍵;⎕IO **
**λ←⍎(⍵>1)⊃'⍵' '⍵×⍙ ⍵-1'⊣⎕IO←0*
In GNU APL a lambda is turned into a normal defined functions right
after its body was parsed.
/// Jürgen
On 08/25/2016 03:15 PM, Louis de Forcrand wrote:
Yep, still 0 after )SIC.
I checked it with this function:
∇F;⎕IO
[1] ⎕IO←~⎕IO ◊ →1∇
And indeed ⎕IO goes back to its original value after )SIC.
Thank you. I should have put that in the original email.
Louis
On 25 Aug 2016, at 04:01, Christian Robert
<christian.rob...@polymtl.ca <mailto:christian.rob...@polymtl.ca>> wrote:
After second though, you may have/had a point.
keep investigating.
Xtian.
On 2016-08-24 21:54, Louis de Forcrand wrote:
You are entirely right, I didn't do that when I reproduced it for
the email. However I believe that I did check quad IO after )SIC the
first time it happened and it was still 0.
Almost like the variables being localised are localised after execution?
I'll check again when I can.
Thanks for pointing that out.
Louis
On 25 Aug 2016, at 03:39, Christian Robert
<christian.rob...@polymtl.ca <mailto:christian.rob...@polymtl.ca>>
wrote:
try ⎕IO
*after* )sic
the fact is that when a function is interruped, you see "by the
last" interrupted function. Until you type )sic or )reset
my 2 cents, as usual, I may be wrong.
Xtian.
On 2016-08-24 21:09, Louis de Forcrand wrote:
∇Z←F N;⎕IO
[1] Z←⍎(N>1)⊃'N' 'N×F N-1'⊣⎕IO←0∇
F 6
720
f←{⍎(⍵>1)⊃'⍵' '⍵×f ⍵-1'⊣⎕IO←0;⎕IO}
f 6
SYNTAX ERROR
λ1[1] λ←⍎(⍵>1)⊃'⍵' '⍵×f ⍵-1'⊣⎕IO←0;⎕IO
^ ^
⎕IO
0
)SIC
⎕IO←1
f←{⍎(⍵>1)⊃'⍵' '⍵×⍙ ⍵-1'⊣⎕IO←0;⎕IO}
f 6
VALUE ERROR
⍵×⍙ ⍵-1
^
⎕IO
0
)SIC
As is shown, system variables such as ⎕IO cannot be localised
in lambdas as they can be in del-fns.
The last example is there because I just thought the order of
execution was intriguing: it seems to show that the error with
local ⎕IO was signalled at the end (left) of the lambda. This
of course might be as it should.
In addition I was wondering if there is a way to refer to the
enclosing lambda / function anonymously, i.e. ∇ in Dyalog.
That is how I first entered the faulty symbol, and found this
potential bug.
Good luck,
Louis
