Roger wrote:
>  The reason why I asked is that the dyad u/ was defined
>  and set in concrete before we really understood rank,

>  I think it was Larry Breed who said that "give me the 
>  symbol for a function and I'll come up with a mnemonic".

>  (It is different from Ken's original reasoning, I think.)

You have a way of saying in 5 words what takes me 500.

>  See the table in Section A of 
>  http://www.jsoftware.com/papers/J1990.htm

Thanks for that, I never knew the rules were spelled out so, particularly
not all in one place.  It would be a good place to start the Wiki page I
suggested, or to quote in the front matter of the new Dictionary (perhaps in
the new ยง1, Alphabet and Spelling).

> if we had any choice (which we don't).
> we would probably use u/ for something else

And what would we select?  I vote for the previous suggestion (I can't find
the reference right now), that / be Pepe's seq adverb, or something like it,
so that:

           seq =: (((&.>)/)(@:(|.@:[ , <@:])))(>@:)  

           require'strings'
           (_2 <\ 'alpha';'beta' ; 'gamma';'delta' ) rplc~ seq 'alpha beta
gamma delta' 
        beta beta delta delta

could be written simply as  rplc~/  (yes, I know rplc already takes multiple
source-target pairs).  And then f/ y  and  x f/ y  would be very obviously
related (/ is the core function of seq).  We could even have a !. for this
so that we got intermediate results, as in the 2 f 3 f 4 f m thread [1].  

-Dan

[1]  Start of 2 f 3 f 4 f m thread:  
     http://www.jsoftware.com/pipermail/programming/2006-May/002227.html

     Or could we just use  x ((f >@{.) ; ])/ y  or something like that to
     get intermediate results without a special  !.  ?

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

Reply via email to