It's better to actually have the reference to the function in the map, rather than a word that refers to the function. If you store a word then you have to make sure the word is bound to an object which has a reference to that function, like this:
handlers: object [
foo-func: func [ a b ] [ print "foo" ]
bar-func: func [ a b ] [ print "bar" ]
]
handler-names: map [
"foo" foo-func
"bar" bar-func
]
apply get in handlers select handler-names name args
But if you just have a reference to the function in your map, you don't have to do the double indirect, and your code looks like this:
handlers: map reduce [
"foo" func [ a b ] [ print "foo" ]
"bar" func [ a b ] [ print "bar" ]
]
apply select handlers name args
Cleaner code, and more efficient too. Or if you're careful enough, like this:
handlers/(name) a b
The path method above will also work if you want the code to do nothing if there is no handler - common in cases where you have optional handlers, such as in GUIs.
You can even have more than one reference to the same function with different key names. You don't have to assign functions to words, they're just values. You can also use the path method to collect the handlers in the first place, saving a reduce
.
handlers: make map! 10 ; preallocate as many entries as you expect
handlers/("foo"): func [ a b ] [ print "foo" ]
handlers/("bar"): func [ a b ] [ print "bar" ]
handlers/("baz"): select handlers "bar" ; multiple references
That path syntax is just another way to call poke
, but some prefer it. We have to put the string values in parens because of a (hopefully temporary) syntax conflict, but within those parens the string keys work. It's a faster alternative to do select
or poke
.