Yes that is it.  

>I think it's confusing because

X (a0 a1)

in this case is not the same as

(X a0) a1



and not the same as (the implied parentheses created that usually work):

X a0 a1


The workarounds are to use a linear representation of the adverb train and 
another adverb that combines it with its argument.  (happy to share if any 
interest)


________________________________
From: Thomas Costigliola <[email protected]>
To: [email protected] 
Sent: Tuesday, March 15, 2016 12:12 PM
Subject: Re: [Jprogramming] Unbox request for requests


I believe this is what Pascal is trying to demonstrate (correct me if I 
am wrong):

    0 (1 : '/') \
/\

    0 ((1 : '/') \)
|syntax error
|       0((1 :'/')\)

A sequence of adverb applications when the first application returns an 
adverb is okay (it produces the adverb train), however, applied as a 
train of adverbs, which one might expect to behave the same, it is a 
syntax error.

I think it's confusing because

X (a0 a1)

in this case is not the same as

(X a0) a1


On 03/13/2016 10:32 AM, Raul Miller wrote:
> As you point out, this works:
>     + - (1 : '`u') `:6
> + -
>
> If we fully parenthesize that expression, it looks like this:
>     require 'trace'
>     paren '+ - (1 : ''`u'') `:6'
> ((+ (- (1  : '`u'))) `: 6)
>
> So when you put parenthesis around (1 : '`u') `:6 you break the expression.
>
> And, as an aside, your original expression did not actually need any
> parenthesis:
>     + - 1 : '`u' `:6
> + -
>
> Thanks,
>

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to