mirror of
https://github.com/adambard/learnxinyminutes-docs.git
synced 2024-12-23 09:41:36 +00:00
12 KiB
12 KiB
| name | filename | contributors | |||
|---|---|---|---|---|---|
| Sing | learnsing.sing |
|
The purpose of sing is to provide a simple, safe, fast language that can be a good replacement for c++ for high performance applications.
Sing is an easy choice because it compiles to human-quality readable c++.
Because of that, if you work for a while with Sing and, at any time, you discover you don't like Sing anymore, you lose nothing of your work because you are left with nice and clean c++ code.
In some way you can also think Sing as a tool to write c++ in a way that enforces some best practices.
/* Multi- line comment.
/* It can be nested */
Use it to remark-out part of the code.
It leaves no trace in the intermediate c++ code.
(sing translates into nice human readable c++)
*/
// Single line comment, can be placed only before a statement or declaration...
// ...or at the right of the first line of a statement or declaration.
// single line comments are kept into c++.
//
// here we declare if we need to use public declarations from other files.
// (in this case from files 'sio', 'sys')
requires "sio";
requires "sys";
//
// A sing function declaration.
// All the declarations can be made public with the 'public' keyword.
// All the declarations start with a keyword specifying the type of declaration
// (in this case fn for function) then follows the name, the arguments and the
// return type.
//
// Each argument starts with a direction qualifyer (in, out, io) which tells if
// the argument is an input, an output or both...
// ...then follows the argument name and the type.
public fn singmain(in argv [*]string) i32
{
// print is from the sio file and sends a string to the console
sio.print("Hello World\n");
// type conversions are allowed in the form of <newtype>(expression).
sio.print(string(sum(5, 10)) + "\n");
// For clarity you can specify after an argument its name separated by ':'.
var result i32;
recursive_power(10:base, 3:exponent, result);
// referred here to avoid a 'not used' error.
learnTypes();
// functions can only return a single value of some basic type.
return(0);
}
// You can have as many arguments as you want, comma separated.
// You can also omit the 'in' direction qualifyer (it is the default).
fn sum(arg1 i32, arg2 i32) i32
{
// as 'fn' declares a function, 'let' declares a constant.
// With constants, if you place an initializer, you can omit the type.
let the_sum = arg1 + arg2;
return(the_sum);
}
// Arguments are passed by reference, which means that in the function body you
// use the argument names to refer to the passed variables.
// Example: all the functions in the recursion stack access the same 'result'
// variable, supplied by the singmain function.
fn recursive_power(base i32, exponent i32, out result i32) void
{
if (exponent == 0) {
result = 1;
} else {
recursive_power(base, exponent - 1, result);
result *= base;
}
}
//**********************************************************
//
// TYPES
//
//**********************************************************
fn learnTypes() void
{
// the var keyword declares mutable variables
// in this case an UTF-8 encoded string
var my_name string;
// ints of 8..64 bits size
var int0 i8;
var int1 i16;
var int2 i32;
var int3 i64;
// uints
var uint0 u8;
var uint1 u16;
var uint2 u32;
var uint3 u64;
// floats
var float0 f32;
var float1 f64;
// complex
var cmplx0 c64;
var cmplx1 c128;
cmplx0 = 0;
cmplx1 = 0;
// and of course...
var bool0 bool;
// type inference: by default constants are i32, f32, c64
let an_int32 = 15;
let a_float32 = 15.0;
let a_complex = 15.0 + 3i;
let a_string = "Hello !";
let a_bool = false;
// To create constant of different types use a conversion-like syntax:
// NOTE: this is NOT a conversion. Just a type specification
let a_float64 = f64(5.6);
// in a type definition [] reads as "array of"
// in the example []i32 => array of i32.
var intarray []i32 = {1, 2, 3};
// You can specify a length, else the length is given by the initializer
// the last initializer is replicated on the extra items
var sizedarray [10]i32 = {1, 2, 3};
// Specify * as the size to get a dynamic array (can change its length)
var dyna_array [*]i32;
// you can append items to a vector invoking a method-like function on it.
dyna_array.push_back(an_int32);
// getting the size of the array. sys.validate() is like assert in c
sys.validate(dyna_array.size() == 1);
// a map that associates a number to a string.
// "map(x)..." reads "map with key of type x and value of type..."
var a_map map(string)i32;
a_map.insert("one", 1);
a_map.insert("two", 2);
a_map.insert("three", 3);
let key = "two";
// note: the second argument of get_safe is the value to be returned
// when the key is not found.
sio.print("\nAnd the value is...: " + string(a_map.get_safe(key, -1)));
// string concatenation
my_name = "a" + "b";
}
// an enum type can only have a value from a discrete set.
// can't be converted to/from int !
enum Stages {first, second, last}
// you can refer to enum values (to assign/compare them)
// specifying both the typename and tagname separated with the '.' operator
var current_stage = Stages.first;
//**********************************************************
//
// POINTERS
//
//**********************************************************
// This is a factory for a dynamic vector.
// In a type declaration '*' reads 'pointer to..'
// so the return type is 'pointer to a vector of i32'
fn vectorFactory(first i32, last i32) *[*]i32
{
var buffer [*]i32;
// fill
for (value in first : last) {
buffer.push_back(value);
}
// The & operator returns the address of the buffer.
// You can only use & on local variables
// As you use & on a variable, that variable is allocated on the HEAP.
return(&buffer);
}
fn usePointers() void
{
var bufferptr = vectorFactory(0, 100);
// you don't need to use the factory pattern to use pointers.
var another_buffer [*]i32;
var another_bufferptr = &another_buffer;
// you can dereference a pointer with the * operator
// sys.validate is an assertion (causes a signal if the argument is false)
sys.validate((*bufferptr)[0] == 0);
/*
// as all the pointers to a variable exit their scope the variable is
// no more accessible and is deleted (freed)
*/
}
//**********************************************************
//
// CLASSES
//
//**********************************************************
// This is a Class. The member variables can be directly initialized here
class AClass {
public:
var public_var = 100; // same as any other variable declaration
fn is_ready() bool; // same as any other function declaration
fn mut finalize() void; // destructor (called on object deletion)
private:
var private_var string;
// Changes the member variables and must be marked as 'mut' (mutable)
fn mut private_fun(errmsg string) void;
}
// How to declare a member function
fn AClass.is_ready() bool
{
// inside a member function, members can be accessed through the
// 'this' keyword and the field selector '.'
return(this.public_var > 10);
}
fn AClass.private_fun(errmsg string) void
{
this.private_var = errmsg;
}
// using a class
fn useAClass() void
{
// in this way you create a variable of type AClass.
var instance AClass;
// then you can access its members through the '.' operator.
if (instance.is_ready()) {
instance.public_var = 0;
}
}
//**********************************************************
//
// INTERFACES
//
//**********************************************************
// You can use polymorphism in sing defining an interface...
interface ExampleInterface {
fn mut eraseAll() void;
fn identify_myself() void;
}
// and then creating classes which implement the interface
// NOTE: you don't need (and cannot) re-declare the interface functions
class Implementer1 : ExampleInterface {
private:
var to_be_erased i32 = 3;
public:
var only_on_impl1 = 0;
}
class Implementer2 : ExampleInterface {
private:
var to_be_erased f32 = 3;
}
fn Implementer1.eraseAll() void
{
this.to_be_erased = 0;
}
fn Implementer1.identify_myself() void
{
sio.print("\nI'm the terrible int eraser !!\n");
}
fn Implementer2.eraseAll() void
{
this.to_be_erased = 0;
}
fn Implementer2.identify_myself() void
{
sio.print("\nI'm the terrible float eraser !!\n");
}
fn interface_casting() i32
{
// upcasting is automatic (es: *Implementer1 to *ExampleInterface)
var concrete Implementer1;
var if_ptr *ExampleInterface = &concrete;
// you can access interface members with (guess what ?) '.'
if_ptr.identify_myself();
// downcasting requires a special construct
// (see also below the conditional structures)
typeswitch(ref = if_ptr) {
case *Implementer1: return(ref.only_on_impl1);
case *Implementer2: {}
default: return(0);
}
return(1);
}
// All the loop types
fn loops() void
{
// while: the condition must be strictly of boolean type
var idx = 0;
while (idx < 10) {
++idx;
}
// for in an integer range. The last value is excluded
// 'it' is local to the loop and must not be previously declared
for (it in 0 : 10) {
}
// reverse direction
for (it in 10 : 0) {
}
// configurable step. The loop stops when it's >= the final value
for (it in 0 : 100 step 3) {
}
// with an auxiliary counter.
// The counter start always at 0 and increments by one at each iteration
for (counter, it in 3450 : 100 step -22) {
}
// value assumes in turn all the values from array
var array [*]i32 = {0, 10, 100, 1000};
for (value in array) {
}
// as before with auxiliary counter
for (counter, value in array) {
}
}
// All the conditional structures
interface intface {}
class c0_test : intface {public: fn c0stuff() void;}
class delegating : intface {}
fn conditionals(in object intface, in objptr *intface) void
{
let condition1 = true;
let condition2 = true;
let condition3 = true;
var value = 30;
// condition1 must be a boolean.
if (condition1) {
++value; // conditioned statement
}
// you can chain conditions with else if
if (condition1) {
++value;
} else if (condition2) {
--value;
}
// a final else runs if any other condition is false
if (condition1) {
++value;
} else if (condition2) {
--value;
} else {
value = 0;
}
// based on the switch value selects a case statement
switch (value) {
case 0: sio.print("value is zero"); // a single statement !
case 1: {} // do nothing
case 2: // falls through
case 3: sio.print("value is more than one");
case 4: { // a block is a single statement !
value = 0;
sio.print("how big !!");
}
default: return; // if no one else matches
}
// similar to a switch but selects a case based on argument type.
// - object must be a function argument of type interface.
// - the case types must be classes implementing the object interface.
// - in each case statement, ref assumes the class type of that case.
typeswitch(ref = object) {
case c0_test: ref.c0stuff();
case delegating: {}
default: return;
}
// - object must be an interface pointer.
// - the case types must be pointers to classes implementing the objptr interface.
// - in each case statement, ref assumes the class pointer type of that case.
typeswitch(ref = objptr) {
case *c0_test: {
ref.c0stuff();
return;
}
case *delegating: {}
default: sio.print("unknown pointer type !!");
}
}
Further Reading
If you want to play with sing you are recommended to download the vscode plugin. Please follow the instructions at Getting Started