package scalite.tutorial package scalite.tutorial
class Point(xc: Int, yc: Int) class Point(xc: Int, yc: Int) {
var x: Int = xc var x: Int = xc
var y: Int = yc var y: Int = yc
def move(dx: Int, dy: Int) = def move(dx: Int, dy: Int) = {
x = x + dx x = x + dx
y = y + dy y = y + dy
}
override def toString() = override def toString() = {
"(" + x + ", " + y + ")" "(" + x + ", " + y + ")"
}
}
object Run object Run {
def apply() = def apply() = {
val pt = new Point(1, 2) val pt = new Point(1, 2)
println(pt) println(pt)
pt.move(10, 10) pt.move(10, 10)
pt.x pt.x
}
}Scalite is an experimental whitespace-delimited syntax for the scala programming language. This lets you delimit block scope using indentation rather than curly braces, reducing the amount of unnecessary curly braces within the source code. This is an important step in view of the great curly-brace shortage of 2007.
You can use Scalite in your own projects via
// project/build.sbt
addSbtPlugin("com.lihaoyi" % "scalite-sbt-plugin" % "0.1.0")
// build.sbt
scalite.SbtPlugin.projectSettings
scalaVersion := "2.11.4"This will cause any .scalite files in your src/main/scalite and src/test/scalite folders to be picked up by the Scalite compiler plugin automatically. Your .scalite files can interop perfectly with existing .scala code, e.g. calling back and forth. Error reporting, incremental compilation, and all that should work great. Note that Scalite only works with Scala 2.11.x.
Scalite blocks are delimited by indentation rather than curly braces. Thus in the following code,
val x = val x = {
val y = 1 val y = 1
val z = 2 val z = 2
y + z y + z
}
var a = var a = {
1 + 2 + 3 1 + 2 + 3
}
def apply() = def apply() = {
x + a x + a
// 9 // 9
}y and z are local variables only scoped to the definition of x, and not visible outside it. The same rule applies for for loops, if/else/while/do/try blocks. Here's some samples from the unit tests:
var x = 0 var x = 0
for(i <- 0 until 10) for(i <- 0 until 10) {
val j = i * 2 val j = i * 2
val k = j + 1 val k = j + 1
x += k x += k
}
val list = val list = {
for(i <- 0 to x) yield for(i <- 0 to x) yield {
val j = i + 1 val j = i + 1
i * j i * j
}
}
list.max list.max
// 10100 // 10100Multi-line for- and for-yield blocks work too:
val all = for val all = for {
x <- 0 to 10 x <- 0 to 10
y <- 0 to 10 y <- 0 to 10
if x + y == 10 if x + y == 10
yield } yield {
val z = x * y val z = x * y
z z
}
all.max all.max
// 25 // 25var x = 0 var x = 0
var y = 0 var y = 0
while (x < 10) while (x < 10) {
if (x % 2 == 0) if (x % 2 == 0) {
x = x + 1 x = x + 1
y += x y += x
else } else {
x = x + 2 x = x + 2
y += x y += x
}
}
y y
// 36 // 36case object ObjectCase case object ObjectCase {
val w = 1 val w = 1
}
object ObjectLol object ObjectLol {
val x = 100 val x = 100
}
trait MyTrait trait MyTrait {
val y = 10 val y = 10
}
class TopLevel extends MyTrait class TopLevel extends MyTrait {
def apply(): String = def apply(): String = {
val z = 1 val z = 1
import ObjectLol._ import ObjectLol._
import ObjectCase._ import ObjectCase._
"Hello World!" + (a + w + x + y + z) "Hello World!" + (a + w + x + y + z)
// Hello World!113 // Hello World!113
}
val a = 1 val a = 1
}Match blocks are similarly indentation delimited, but with a twist: since there isn't any ambiguity between more case clauses and statements outside the match block, Scalite does not require you to indent the case clauses, saving you one level of indentation:
val z = (1 + 1) match val z = 1 match {
case 1 => case 1 =>
println("One!") println("One!")
"1" "1"
case 2 => "2" case 2 => "2"
}
z z
// "2" // "2"The same applies to the catch block of a try-catch expression:
try try {
println("Trying...") println("Trying...")
x.toString x.toString
catch } catch {
case n: NullPointerException => case n: NullPointerException =>
println("Dammit") println("Dammit")
"null" "null"
}In addition to indentation-scoped blocks, for/if/while blocks also support a paren-less syntax if the generators of the for or the conditional of the if or while fit on a single line:
var x = 0 var x = 0
for i <- 0 until 10 for (i <- 0 until 10) {
val j = i * 2 val j = i * 2
val k = j + 1 val k = j + 1
x += k x += k
}
val list = val list = {
for i <- 0 to x yield for i <- 0 to x yield {
val j = i + 1 val j = i + 1
i * j i * j
}
}
list.max list.max
// 10100 // 10100
var x = 0 var x = 0
var y = 0 var y = 0
while x < 10 while (x < 10) {
if x % 2 == 0 if (x % 2 == 0) {
x = x + 1 x = x + 1
y += x y += x
else } else {
x = x + 2 x = x + 2
y += x y += x
}
}
y y
// 36 // 36You can use whitespace-delimited blocks for your own functions too, and not just for built-in control flow constructs, using the do keyword:
val xs = 0 until 10 val xs = 0 until 10
val ys = xs.map do val ys = xs.map{
x => x + 1 x => x + 1
}
ys.sum ys.sum
// 55 // 55
val zs = xs.map do val zs = xs.map{
case 1 => 1 case 1 => 1
case 2 => 2 case 2 => 2
case x if x % 2 == 0 => x + 1 case x if x % 2 == 0 => x + 1
case x if x % 2 != 0 => x - 1 case x if x % 2 != 0 => x - 1
}
zs.sum zs.sum
// 45 // 45The do at the end of the can be made optional with a slightly cleverer parser, but for now it is required.
You can also use the do with a function that takes an argument, like this:
val ws = xs.map do x => val ws = xs.map { x =>
val x1 = x + 1 val x1 = x + 1
x1 * x1 x1 * x1
}
ws.sum ws.sum
// 385 // 385Scalite supports spreading out the header of a for-loop over multiple lines:
val all = for val all = for {
x <- 0 to 10 x <- 0 to 10
y <- 0 to 10 y <- 0 to 10
if x + y == 10 if x + y == 10
yield } yield {
val z = x * y val z = x * y
z z
}
all.max all.max
// 25 // 25
var i = 0 var i = 0
for for {
x <- 0 to 10 x <- 0 to 10
y <- 0 to 10 y <- 0 to 10
if x + y == 10 if x + y == 10
do } {
val z = x * y val z = x * y
i += z i += z
}
i i
// 165 // 165As well as the conditional of an if-statement:
if if ({
println("checking...") println("checking...")
var j = i + 1 var j = i + 1
j < 10 j < 10
do }) {
println("small") println("small")
1 1
else } else {
println("big") println("big")
100 100
}Or while loop:
var i = 0 var i = 0
var k = 0 var k = 0
while while({
println("Check!") println("Check!")
var j = i + 1 var j = i + 1
j < 10 j < 10
do }){
println("Loop!") println("Loop!")
i += 1 i += 1
k += i k += i
}
k k
// 45 // 45As you can see, the do keyword is used to indicate that the previous block has ended and a new block begins, in situations where in the default Scala syntax you only have a }{ or ){ to separate these expressions. There should be no ambiguity with a do-while loop due to the fact that the do-while and while-do/for-do/if-dp always come together.
Despite the fact that do-while loops still work, they are rendered redundant by the fact that the condition and body of the while loop are now symmetrical: Both sections of the loop can now easily hold an arbitrary block of statements, and any statements that you wish to execute before the condition is checked for the first time can simply be placed in the uppder block before the do.
Any do-while loop of the form
do {
A
} while (B)Can be equivalently rewritten as
while
A
B
()With a slightly cleverer parser, the ugly () at the end of the loop can be removed, although for now it is required.
Want to see more? I have ported uPickle's pure-scala JSON parser to Scalite! Comparing the original code with the Scalite version should give a good sense of what a messy, real world code base looks like in Scalite. You can also leaf through the unit tests samples for more examples of what Scalite code looks like.
Read previous discussion on the Google Group if you want to know the background behind this.
Scalite implements approximately the following transform to convert the Scalite code to valid Scala:
- If the last tree T on a line is a class/trait/object-header, def-header, var/val/lazyval, do, or control-flow construct
- And it is immediately followed by a
\nwith no{ - Then insert a
{at the end of that line, - And insert a
}at the beginning of the first line whose indentation is less-than-or-equal to the indentation of the line of the start of the tree T and the first token of that line is not acase
Although there are a myriad of edge cases in this translation, this is the approximate algorithm that should explain the bulk of Scalite's behavior.
Notably, the fact that Scalite only special-cases lines which do not end in a { means that old-fashioned curly-brace Scala continues to function perfectly fine, and can be mixed together with Scalite code in the same source files and still compile without issue.
Rather than being a hacky text-manipulator, Scalite is implemented as a modification to the Scala compiler that performs the transformation directly on the token-stream being produced by Scala's lexer. Scalite also has access to Scala's parser, which lets it recognize language constructs on an AST (and not just the token) level and hopefully provide a more robust implementation of the whitespace-delimited syntax.
A more robust solution would be to fork the Scala compiler's recursive-descent parser, in order to properly insert the modifications in the correct places in the grammar. This would require forking the entire compiler code base, as the Scala compiler and parser have a circular type-level dependency on each other, and the parser cannot be easily swapped out. Although possible, this approach was more work than I was willing to put in.
Scalite is implemented as a custom Global rather than as a compiler plugin, because the Scala compiler architecture does not allow compiler plugins to replace the lexing-and-parsing phase of the compilation pipeline. This makes it difficult to bundle up and re-use in other projects. Nevertheless, there is a moderately large suite of unit tests which uses this custom Global to programmatically compile chunks of code and executes them to ensure they behave as expected.
Scalite is the culmination of about 30 hours of work, and isn't ready to be used for anything at all. The semantics are full of bugs, and the implementation is a rats nest of complexity, but it works, and hopefully will inspire or convince someone somewhere that a whitespace-based syntax is something worth trying out.