Henry wrote:
>...
>line_producing_the_desired_result
>if. unmap_jmf_... do. end.
>)
Yes, at one point I had:
line_producing_the_desired_result
assert. 1 [ unmap_jmf_ ...
But I didn't like it, and since it added a line to my definition anyway, I
opted for the more obvious and readable IP=. ... .
By the way, here's something you might find cute. Remember the diamond symbol
from APL? There's an equivalent in J:
3 : ' ''execute'' label_. ''from'' label_. ''left to right'' ' 0
left to right
Dependable, and no side effects. Won't help with the "unmap needs to be called
after the result is calculated, but the result calculated is the result
returned" situation, but might be useful in others.
Raul wrote:
> Given that the dictionary defines an order of execution
First, leveraging that order demands requires care and forethought [1], which
seems a large price to avoid using an extra line (in particular for code
maintenance, as I might change the ultimate line without bearing in mind the
subtleties of order-of-execution).
Second, The Dictionary is subject to change without notice.
-Dan
[1] Fun with order of execution:
0 0$(smoutput 'first')[smoutput 'second'
first
second
0 0$(smoutput 'first')[(smoutput 'second')
second
first
0 0$(smoutput 'first')[ ASSIGNMENT=:smoutput 'second'
second
first
0 0$(smoutput 'first') ] smoutput 'second'
first
second
NB. Ok, so parens get done first...
NB.
NB. Which means I can do an assignment in the
NB. parens and use the assignment later in the line...
0 0$(smoutput A=:'first') ] smoutput '2: ', A
|domain error
| 0 0$(smoutput A=:'first')]smoutput'2: ' ,A
NB. Huh. Guess not.
NB. Check this one out:
(smoutput 'first') 1 :'u' smoutput 'second'
first
NB. Theoretically, the assignment is idempotent wrt. the
NB. result of the line (i.e. it's just a side-effect)...
(smoutput 'first') 1 :'u' ASSIGNMENT=:smoutput 'second'
second
first
|syntax error
| (smoutput'first')1 :'u'ASSIGNMENT=:smoutput'second'
All of which are just curiosities, unless you're depending upon side effects.
Which, of course, is exactly what we're doing in the "unmap" situation.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
