My initial reaction to this change was to favour it. Oleg's message
expresses much better than I could its potential advantages.
However it does impose a new discipline in thinking about names in
definitions - it is essential to recognise that the first occurence of an x
or y (or m n u v) in an assignment obliterates the function argument.
A very powerful feature of J is that it permits construction and execution
of code from function arguments. With this change it is clearly dangerous
to
execute such code before all references to the function arguments are
complete
or have been protected by saving the argument values.
The following simple example of test illustrates a case where both x and y
are recoded by ". even though there is no apparent reference to y on the
left of a copula - the reference is in the data supplied argument.
test =: 4 : 0
". '''',x,'''=.y'
x;y; ". each ;:x
)
'a b c' test i. 3 5
+-----+--------------+---------+---------+--------------+
|a b c| 0 1 2 3 4|0 1 2 3 4|5 6 7 8 9|10 11 12 13 14|
| | 5 6 7 8 9| | | |
| |10 11 12 13 14| | | |
+-----+--------------+---------+---------+--------------+
'x y z' test i. 3 5
|domain error: test
| x;y;".each ;:x
test1 =: 4 : 0 NB. uses inflected names
". '''',x.,'''=.y.'
x.;y.; ". each ;:x.
)
'a b c' test1 i. 3 5
+-----+--------------+---------+---------+--------------+
|a b c| 0 1 2 3 4|0 1 2 3 4|5 6 7 8 9|10 11 12 13 14|
| | 5 6 7 8 9| | | |
| |10 11 12 13 14| | | |
+-----+--------------+---------+---------+--------------+
'x y z' test1 i. 3 5
+-----+--------------+---------+---------+--------------+
|x y z| 0 1 2 3 4|0 1 2 3 4|5 6 7 8 9|10 11 12 13 14|
| | 5 6 7 8 9| | | |
| |10 11 12 13 14| | | |
+-----+--------------+---------+---------+--------------+
The function test1 with the inflected form creates no difficulty. I am
afraid I very much dislike code which has unintended side effects dependent
on data provided at execution time.
This change now means we have two classes of nouns. Those which we can code
as we always have, and the small group of single character names where we
have to explicitly consider possible side effects from their use as
placeholders as well.
I am not against getting rid of the inflection for placeholders - an
alternative I have used is the conjunction 'of' in
system\packages\misc\fndef which permits definitions of the form
trianglearea =: function of 'height base'
0.5*base*height
)
Newbies and young users can understand that form of definition very easily,
but it is much simpler to code when the place holders are inflected names
and thereby distinguished from the usual alphameric set of variable names.
Fraser
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
