Functional programming

[Chris Angelico]

On Tue, Mar 4, 2014 at 1:38 AM, Rustom Mody <rustompmody at gmail.com> wrote:
> If you want the (semantic) equivalent of python's [1,2,'foo']
> you need to make an explicit union Int and String and its that
> *single* union type's elements that must go in.
>
> In all cases its always a single type. And so
> sum([1,2,[3])

Okay. That's how the declaration goes, then. So how do you tell it
that 1 isn't an Int, it's a member of the union of Int and String? How
do you create a list which has [Int_String(1), Int_String(2)] and is
therefore allowed to be added to [Int_String('foo')] ? Can you do that
with literals?

This is why it's tricky to put rules in based on type inference. The
programmer's intent isn't in the picture. If Python ever acquires that
kind of restriction ("here's a list that can contain only this type /
these types of object"), I would hope that it's left up to the
programmer, not the compiler, to stipulate. That's how it is with Pike
(if you just say "array", it can take anything), and that's the only
way to be sure the programmer doesn't have to fight the language.

You said earlier

>> On Tue, Mar 4, 2014 at 1:08 AM, Rustom Mody wrote:
>> > If 'integer-less-than-3' were a type then yes there would be this
>> > problem. More generally, if types could overlap then automatic
>> > type-inference is impossible

The type "Int" overlaps with the type "Union of Int and String". How
is that resolved? Type inference ignores unions? That's the only way I
can think of. Hence the original difficulty of type-inferring on a
list that isn't complete yet.

ChrisA