Version 7 (modified by simonpj, 7 years ago) (diff)


GHC Status May 2008

The last six months have been a time of consolidation for GHC. We have done many of the things described in the last HCAR status report (November 2007), but there are few new headline items to report, so this status report is briefer than usual.

Highlights of the last six months

  • Simple language extensions
    • Record syntax: wild-card patterns, punning, and field disambiguation
    • View patterns
    • Generalised list comprehensions
    • Quasi-quoting
  • Type-indexed families. We learned a lot by writing a paper about the question of type inference in the presence of type families (and existentials, and GADTs): "Type checking with open type functions". The implementation has not quite caught up with the paper and is still incomplete in many ways, but it's a focus of active work.
  • Impredicative polymorphism. We are not happy with GHC's current implementation of impredicative polymorphism, which is rather complicated and ad hoc. Dimitrios (with Simon and Stephanie) wrote a paper about a new and better approach: "FPH : First-class Polymorphism for Haskell". At the same time, Daan Leijen has been working on his closely-related design: "Flexible types: robust type inference for first-class polymorphism". Daan's design has a much simpler implementation, in exchange for an (arguably) less-predictable specification. Which of these two should we implement? Let us know!
  • External Core. Tim Chevalier has been working on making GHC emit External Core that can be correctly parsed, typechecked, and interpreted by a separate Haskell program. This isn't the same as the ability to round-trip External Core back into GHC, but it's a big step forward from the bit-rotted state into which External Core had fallen.

Nested data parallelism

We have been working hard on Data Parallel Haskell, especially Roman Leshchinskiy and Gabriele Keller. It has turned out be be hard to get the entire transformation and optimisation stack to work smoothly, and we have not made progress announcements because we don't want to yell about it until it Actually Works. But it is the biggest single GHC focus: Roman works on it full time.

Manuel: more to say.

We expect to release a working version of Data Parallel Haskell as part of GHC 6.10 (see below).

Other current activities

  • Work on the back end has been stalled, but John Dias started a 6-month internship in April, so expect progress on this front.
  • Thomas Schilling is doing a Google Summer of Code project to improve the GHC API.

  • Max Bolingbroke is doing a Google Summer of Code project to make it easy to build a plug-in for GHC; for example, a new optimisation or analysis pass.

Release plans

We plan to release GHC 6.8.3 at the end of May 2008, with many bug-fixes but no new features.

We plan to release GHC 6.10 around the time of ICFP, with significant new features. The up-to-date list of new stuff is kept at, but here's a quick summary:

  • Simple language extensions (mentioned above)
  • Type-indexed families
  • Data Parallel Haskell
  • Parallel garbage collection
  • Extensible exceptions
  • External Core
  • Shared libraries
  • Improved back end
  • Further library reorganisation