The Scala Compiler Corner:     A collection of resources for compiler hackers Companion websites: All about the Scala.NET compiler The jdk2ikvm automatic migration tool

Code walkthroughs of the compiler

• First-half of the pipeline: parser, namer, packageobjects, typer, superaccessors, pickler, refchecks, liftcode, uncurry, tailcalls, specialize
• Second-half of the pipeline: explicitouter, AddInterfaces, erasure, lazyvals, lambdalift, constructors, flatten, mixin, cleanup
• ICode and its optimizers: GenICode, inliner, inlineExceptionHandlers, closelim, dce
• Backends: GenASM, GenJVM, GenMSIL.

Compiler plugins

• Experimental optimizations
  • Parallelization of optimization phases (January 2012)
  • Transforming Function nodes (standing for closures) before uncurry turns them into anonymous classes (November 2011)
  • Partial evaluation based on three-address-form, including conversion back to stack-based language (November 2011)
  • Moving Scala ASTs one step closer to C, by turning them into three-address form (a Jimple-like IR) (September 2011)
  
  
• Miscellaneous AST transformation
  • Work in progress: GOTO elimination in Scala ASTs (February 2011)
  • Work in progress: What about an explicit formulation of module initialization as ICode-level desugarings (May 2010)
  
  
• Source-level, AST-backed transformations that preserve layout
  • Finding the source code fragment that corresponds to a given AST-level type-reference (January 2011)
  • A source-level, automatic API migration that preserves layout (a story of range positions) (January 2011)
  
  
• Unparsing
  • Work in progress: Unparsing types the Scaladoc way (January 2011)
  • Bits and pieces of information about the parser, namer, and typer phases that turn out to be necessary just to be able to unparse Scala ASTs (December 2010)
  • Learning and doing scalac transformations the easy way: via unparsing (December 2010)

ICode transformation

• Dataflow Analyses on ICode (November 2011)
• Using reaching-defs and type-flow analyses to turn ICode into a stackless IR (April 2010). Prototypes: first, second
• SSA, as part of Scala on LLVM by Geoff Reedy

Exploring invokedynamic and MethodHandle

• Runtime metaprogramming via java.lang.invoke.MethodHandle (May 2012)

Related resources

• Talks
• 2012-05-01: Functionality ready to land in optimizer and backend
• 2012-04-18: What the optimizer does to your code (slides, video) at Scala Days 2012
• 2011-11-22: Just trying to generate faster code faster under -optimise
• 2011-02-22: A few additional AST lowerings that can simplify the development of new backends (LaTeX sources)
• 2011-01-18: jdk2ikvm and scala.tools.unparse
• Kiama, a DSL and library for language processing, http://code.google.com/p/kiama/
• Report by Mirko Stocker on the Scala Refactoring Project, http://scala.ifs.hsr.ch/
• Project ideas (backend-independent)
• Emitting Android .dex classes directly.
• Translation of dynamically-typed languages into Scala. PDF
• Typestate analysis for Scala: It would be interesting to know whether declarative points-to analyses (paper and source code) can scale up when object abstractions track typestates. There are different expressiveness levels for typestate properties, all sharing the limitation to temporal safety (no liveness). Recent work includes analyses for socket-style APIs, as well as for more elaborate API protocols (e.g., JDBC). More related work: (1) certain typestate properties are captured by type systems (such as X10's constrained types) supporting property-dependency where "property" means "final value-typed field set at the latest by object instantiation time". (2) The original GoF State design pattern also detects protocol violations, only that at runtime. (3) There are heroic efforts to encode some form of dependent types in Scala (HList and all that, made possible by Yrecursion and experimental). A dedicated compiler plugin can instead dispense with all the resulting syntactic burden. (4) Ordered Datalog for refactoring analyses.