Pride programming language ideas and samples

gistfile1.ooc

// NO TYPES SEEN, NO HARM DONE <- todo :P

// One-liner comment
/*
Multiligne
comment
*/
var :: Int // Declaration
var = 42 // Assignement
blah :: Int = 42 // Wow! :D
foo = 42 // Decl-assign (type infered)

(a :: Int, b :: Int) -> a+b // Function (lambda) literal = [declaration tuple] -> [Instruction Block]

f = (a :: Int, b  :: Float) -> a+b
f (1,41) // Function call = [Function object][argument tuple]

{a = 1, b :: Float, c = "lol"} // Object literal
obj :: {a :: Int, b :: Float, c :: String} // Type is similar to the object itself
obj = {a = 1, b :: Float, c = "lol"}
obj b = 41.0 // Object access

    doStuff()
    returnSomething() // Instruction block -- doStuff will be executed and the block will be evaluated to the return of returnSomething()

obj a = 41
    b = 1.0 // Chain calling

/*
This is not exactly chain calling.
It is actually an instruction block inside obj's scope.
*/

type Foo = {a :: Int, b :: Float, c :: String} // Type alias
foo :: Foo
foo a = 1
foo b = 41.0
foo c = "bar"

// a :: B form
// a :: B is a definition when it is a statement in an instruction block
// Else, it is a cast.
// Let's see a special case
func = -> a = 1
          a :: Float
/*
According to the above rules, you would probably think that a :: Float is a definition and that you would have a redifinition error.
But this is not the case.
The a :: Float statement is read as return a :: Float by the compiler, as it is the last expression in an instruction block.
Thus, the second rule applies and a is casted to a Float and then returned.
*/

[1,2,3,4,5,6,7,8,9] // Array literal
[1..10] // Ranges -> [1,2,3,4,5,6,7,8,9]

// Flow control
// if statement
if true then doTrue() else doFalse() 
// Else can be ignored
if true then doTrue()
/* else if is actually else (if ... )
In general,
if [condition] then [instrBlock1] {else [instrBlock2]}

while statement
*/
while true do stuff()
// or
do stuff() while true
/*
Go read a C manual if you need an explanation for the difference between the two.
In general,
while [condition] do [instrBlock]
or
do [instrBlock] while [condition]

for statement
*/
for i in [1..10] do stuff(i)
// or
for i :: Float in [1..10] do stuffFloat(i)
/*
In general,
for [varName] in [array] do [instrBlock]
or
for [declaration] in [array] do [instrBlock]
*/

(1,41.0,"lol") // Tuple
// :: (Int,Float,String)
// Here is a nice example with tuples
f = -> (1,41.0,"lol")
(_,_,x) = f() // Yay! Kinda tuple retrospecting

// A kind of pattern matching
// Here is the pattern matching literal
(Int,Int) =>
    (0,_) -> 0
    (_,0) -> 0
    (a,b) -> a+b

// With objects
type Foo = { a :: Int, b :: Float }
(Int,Foo) =>
    (0,_) -> 0
    (_,{ a = 0, b = _ }) -> 0
    (a,b) -> a+(b a)+(b b)

// Or
(Int,Foo) =>
    (0,_) -> 0
    (_,{ a = 0 }) -> 0
    (a,b) -> a+(b a)+(b b)
// When we only ask for some members of the object to be the same, the rest are not taken into account

// What is the type of a function?
f :: () -> // A function that takes no argument and returns nothing
g :: () -> Int // A function that takes no argument and returns an Int
h :: (Int,Float) -> Float // A function that takes an Int and a Float argument and returns a Float
/*
A little history? Why () -> ?
If the arrow was omitted, the type would be () (the empty tuple) wich is equivalent to void.
Also, I did not want to have a prefix like Func (see ooc ;) )
That are the deep reasons of doom ^.^
*/

use something // will search in path for something.pri and include it
link something // will cause the C compiler to link against -lsomething

// Here is a sample from the true SDK to show the C FFI
type Int = ctype int with {
   sin :: extern(sin) ()->Int
}