Very often I find myself in a situation like this:
IN: scratchpad: TUPLE: box
length width height ;
IN: scratchpad { { 1 2 3 } { 4 5 6 } { 6 7 8 } }
--- Data stack:
{ ~array~ ~array~ ~array~ }
IN: scratchpad [ V{ } clone-like ] dup [ map ] dip call
--- Data stack:
V{ ~vector~ ~vector~ ~vector~ }
IN: scratchpad [ dup pop swap dup pop swap dup pop swap drop box boa ] map
--- Data stack:
V{ ~box~ ~box~ ~box~ }
IN: scratchpad dup .
V{
T{ box { length 3 } { width 2 } { height 1 } }
T{ box { length 6 } { width 5 } { height 4 } }
T{ box { length 8 } { width 7 } { height 6 } }
}
Which achieves the result I want, but this:
dup pop swap
Must be copy/pasted the same number of times as items in the array I want to destruct, and the array must first be clone-like
d to a V{ }
vector, and... it's just really terrible, messy code. (Note that 3 [ dup pop swap ] times
will not work because of the stack effect.)
There must be a better way to construct an instance of a TUPLE
from the items of an array. What is it?
Note you can unpack many elements at a time from sequences with first2
, first3
and first4
in sequences
and x firstn
from sequences.generalizations
.
I think most "Factorish" way of doing it is defining a constructor that takes a sequence as argument:
: seq>box ( s -- box ) first3 box boa ;
Then you can map
on it:
{ { 1 2 3 } { 4 5 6 } } [ seq>box ] map
Other way, closer to what you're doing:
{ { 1 2 3 } { 4 5 6 } } [ box prefix >tuple ] map
>tuple
takes a sequence of the form
{ name-of-tuple-class 1st-slot-value 2nd-slot-value ... last-slot-value }
so you have to prefix
each sequence with the class.