paulp

(declare-fun a () Int)
(declare-fun b () Int)
(declare-fun c () Int)

;; (assert (= a (div -7 -3)))
;; (assert (= b (div  7 -3)))
;; (assert (= c (div -7  3)))
;; (assert (= a (mod -7 -3)))
;; (assert (= b (mod  7 -3)))
;; (assert (= c (mod -7  3)))

paulp

package object types {

  import scala.language.reflectiveCalls
  import scala.language.higherKinds

  // quantifiers aka (co)ends
  type Forall[+F[_]] = { def apply[X]: F[X] }
  type Exists[+F[_]] = F[_]

  // basic categorical notions

paulp

• Thesis: Problem is the language as much as the implementation of the compiler
  • Big problem is cyclic dependency in the typer where type inference and type-checking and implicit search mutually depend on each other. In other words: Can you write an interpreter without also writing a typer that at least implements type inference and implicit search. Can you do implicit search without implementing all of the type checker? Type inference and implicit search must be disentangled. They know that in scala too, except that it will be close to impossible. Note that disentanglement doesn't preclude interaction, but it has to flow through much narrower, explicit, disciplined channels.
  • Consequence if this is true: If you want to simplify the compiler you would have to identify how to simplify the language to break up those cycles **There are innumerable simplifications available in the compiler which require nothing but sane software engineering. If you want to minimize the ESSENTIAL complexity

paulp

package tuple

import language.experimental.macros
import language.dynamics
import scala.reflect.macros.Context

sealed trait ~[+A, +B]

private[tuple] object TupleMacros {

paulp

// Say we have a List of names and we would like to find all those names where "am" occurs:
{
  // LINQ
  // string[] names = { "Sam", "Pamela", "Dave", "Pascal", "Erik" };
  // List<string> filteredNames = names.Where(c => c.Contains("am"))
  //                                   .ToList();

  // Java Streams
  // String[] names = {"Sam","Pamela", "Dave", "Pascal", "Erik"};
  // List<String> filteredNames = stream(names)

paulp

De-virtualized Collections

De-virtualizing abstract collections would increase performance and decrease bytecode bloat for common use cases–without tying the hands of implementors. Conceptually, operations on abstract collections become syntactic sugar for inlined code templates. You opt-in to dynamic behavior by choosing subtypes that override extension methods with virtual ones. The root collections then become an encapsulation of their elements, but not an abstraction over operations on those elements; you can still perform such operations on these collections, but the compiler statically binds the implementation rather than indirecting to the run-time type's vtable-provided version. Collection sub-families–parallel, non-strict, and the like–override generic extension methods with virtual ones, and any value statically typed to such a collection works in the usual object-oriented way.

Sample de-virtualized collection hierarchy

import collection.generic.CanBuildFrom
import language.im

paulp

De-virtualized Collections

De-virtualizing abstract collections would increase performance and decrease bytecode bloat for common use cases–without tying the hands of implementors. Conceptually, operations on abstract collections become syntactic sugar for inlined code templates. You opt-in to dynamic behavior by choosing subtypes that override extension methods with virtual ones. The root collections then become an encapsulation of their elements, but not an abstraction over operations on those elements; you can still perform such operations on these collections, but the compiler statically binds the implementation rather than indirecting to the run-time type's vtable-provided version. Collection sub-families–parallel, non-strict, and the like–override generic extension methods with virtual ones, and any value statically typed to such a collection works in the usual object-oriented way.

Sample de-virtualized collection hierarchy

import collection.generic.CanBuildFrom
import language.im

paulp

#!/usr/bin/env bash
#
# ^-- don't assume bash is in /bin

# we need a git library of stuff like this, since there are lots
# of ways to get it subtly wrong
git_branch_exists () {
  git show-ref --verify --quiet refs/heads/$1
}

paulp

CmdUtils.CreateCommand({
  names: ["java"],
  arguments: [{role: "object",
               nountype: noun_arb_text,
               label: "search criteria"}],
  icon: "http://www.sun.com/favicon.ico",
  preview: "Searches The Java 6 API.",
  help: "Enter the name of the Java class or package for which you would like to see the documentation.",
  author: {name: "KS", email: "tss@cornbread.com"},

paulp

trait Cloneable[+T <: AnyRef] extends AnyRef {
  def create(): T
  override def clone(): T = create()
}

class A(var a: Int) extends Cloneable[A] {
  def create(): A = new A(a)
}

class B(aa: Int, var b: Int) extends A(aa) with Cloneable[B] {