Implement Let block

[wende]

Add:
Let

Example:

let 
  a = 1
in
  a * 2

->

a = 1
a * 2

[net]

Is there a need for let instead of simply writing

a = 1
a * 2

?

[OvermindDL1]

Is there a need for let instead of simply writing

Beside that it is not valid elm? It allows you to define recursive structures too.

[wende]

Exactly. So far we're trying to be syntactically idiomatic to Elm. When we succeed to copy the whole language we might discuss some changes. But it's definitely not a time for that yet :)

Plus having let... in has a very important implication which is being consistent about functions being only one expression

[net]

It allows you to define recursive structures too.

Could you elaborate on this @OvermindDL1?

[net]

Plus having let... in has a very important implication which is being consistent about functions being only one expression

Assignment is not an expression in Elm.

[OvermindDL1]

Could you elaborate on this @OvermindDL1?

Some valid elm, you can create inline mutually recursive functions and calls and such.

import Html exposing (text)

main =
  let
    r = c 1
    a b = 21 * b
    c d = a (d+1)
  in text (toString r)

I still prefer the OCaml method though, much more clear and redefinable...

[net]

You could still achieve that without the let syntax.

main =
  r = c 1
  a b = 21 * b
  c d = a (d+1)

  text (toString r)

[OvermindDL1]

The purpose is to allow rebinding as well (this is not valid elm because elm is a bit stupid in how it does a lot of things):

import Html exposing (text)

main =
  let
    r = c 1
    a b = 21 * b
    c d = a (d+1) in
  let r = r + 1 in
  text (toString r)

In addition to controlling scoping and more things.

Plus, it does not matter, it is still not valid elm to leave the let out. ;-)

[wende]

Assignment is not an expression in Elm.

@net Exactly. But the let..in block is

[net]

@OvermindDL1 Is rebinding in an expression really that important? Many would argue that rebinding only causes confusion. Pick better variable names instead.

Scoping is entirely possible with the non-let syntax.

[net]

@wende Right, so allowing assignments before an expression does not violate the one-expression-per-function rule.

[OvermindDL1]

@wende Right, so allowing assignments before an expression does not violate the one-expression-per-function rule.

Unless you are getting into whitespace sensitivity (ew) then it becomes ambiguous:

  a = 4
  b = 4 + a
  b + 6

Unless the language is whitespace sensitive (big ew) it is parsed like:

  (a = 4) (b = (4 + (a b) + 6)

The let ... in ... defines bounds.

But still, it is a moot point, elm does not allow it. ;-)

[wende]

@OvermindDL1
Elm is a little bit whitespace sensitive when it comes to case branches and function calls.
For example

...
Something -> 
  fun 
     a
     b
     c
SomethingElse b c ->
...

You can't really know where fun ends, and SomethingElse starts. It's a big issue to solve for us right now #40 and it's gonna be really hard because of whitespace sensitivity.
So I definitely preach to your ew

[OvermindDL1]

Elm is a little bit whitespace sensitive when it comes to case branches and function calls.

I know, it endlessly bugs me. ^.^;

Honestly I'd say you should write a new backend for OCaml that compiles to Elixir instead. ^.^
I've been wanting to do that, but so very much lacking on time...