The 2006 GHC hackathon: 14,15 September 2006, Portland
GHC is used by lots of people, but its implementation is rather over-centralised, even though GHC is a BSD-licensed, open-source project. The biggest obstacle to more people getting involved is that GHC is a big, and hence intimidating, system. The purpose of the GHC hackathon is to give a tutorial in GHC’s innards, plus time to work on some projects in a context where there are plenty of people around to help. The objective is to substantially broaden (more numbers) and deepen (more confidence) the community of people who feel they know enough to fix and enhance GHC.
These days, GHC is more than just a batch compiler:
It's an interactive interpreter
It's a (Haskell) library
The latter, in particular, opens up new possiblities, such as using GHC to parse and type-check Haskell, before analysing or transforming it with your own program. However, GHC-as-a-library has a pretty big interface that uses a lot of data types, which will be among the things we'll describe at the hackathon.
We'll suggest some projects, but you're welcome to come along with your own wacky ideas or itches that you want to scratch, and work on them in an environment with a high-bandwidth connection to developers who really know the innards of GHC. You could work on something individual, or in small groups. We anticipate all being in one room, or in a tight group of rooms, so there’d be lots of informal interaction.
One full day of tutorial from Simon PJ & Simon M about GHC’s glorious innards (14th)
Then one or two days of hacking on projects (15th, 16th)
We expect that some of you will come with GHC-related projects that you want to work on, but we'll also offer a list of possible projects, so you can come without anything in mind and choose on the day. The list is a wiki; please add your own suggestions to it, even if you can't come!
The Hackathon will be hosted by Galois and PSU, and will be held on the PSU campus, in the Fourth Avenue Building, 1900 SW 4th Avenue, Room 10,
near the other ICFP activities. (Note this is changed from the original planned location, near the Galois offices in Beaverton).
Entry to the Forth Avenue Building with be via the main entrance, on the south west corner of the building.
The room has desks, a projector, whiteboards, and wireless access. Bring your own laptop
(capable of building GHC).
Have a look at the TriMet web site for details about how to use Public transportation in Portland.
The Hackathon will be open
9 till 5 - Thur 14th - Breakfast, Lunch and afternoon break snacks.
9 till 5 - Fri 15th - Breakfast, (out for lunch), afternoon break snacks.
9 till 12 - Sat 16th - Breakfast, snacks.
If you want to come, we strongly encourage you to tell us, in two ways:
Email jodee@…, stating you wish to attend the GHC Hackathon.
Add your name to the list of participants below (this page is a Wiki, so you can do that yourself).
Registration is not absolutely required, but any Pizza, manuals, and workspace will be allocated on a first come, first served basis, so registration is a good plan!
Zero. But you may have to feed yourself if you don't register.
Areas of Interest
[Ext] - Extending Haskell; things like Template Haskell, and new syntax.
[TC] - Typechecking.
[Opt] - Optimizing Haskell, GHC Core representation, and tools to help optimization.
[CG] - Turning GHC Core Haskell into C, or Java, or C#, or C--, cross-compiling.
[i] - Debugging and GHCi, and other interactive tools.
[Lib] - Libraries, packaging, and namespace issues.
We had some enquiries about whether people should come along with a
fresh GHC build ready to go, and if so how to go about getting one.
I'll probably be demonstrating how to set up a build on Thursday, so if
you want you can wait until then and there should be plenty of time to
get a working build before Friday. It's probably a good idea to make
sure that your machine has the relevant tools, though:
In particular, the section "Quick start for developers" should have
enough to set you up with a build that won't take several days to
complete, and will be tailored to what you're particularly interested in
There are two branches of the source code now: the HEAD and the 6.6
branch. They are still very close, so building either one should be
fine. The 6.6 branch darcs repositories are all under
If you want to come, please add your name to the list below. Add a sentence or two about yourself and why you want to come. If you are working on a project to do with GHC, tell us about that too.
Simon Marlow One of the main authors of GHC, particularly the back end, run-time system, and libraries.
Simon Peyton Jones One of the main authors of GHC, particularly the type system and Core intermediate language.
Andy Adams-Moran, interested in making cross-compiling with GHC easier, and in learning more about the innards of the compiler (versus the RTS)
Björn Bringert, Chalmers University of Technology and Göteborg University, want to learn about the internals of GHC for future projects, possibly implement instance deriving for GADTs
Karl Crary, CMU, working on a Haskell-driven typesetting system
Iavor S. Diatchki, interested in Haskell for systems programming/type checking (general improvement rules)
Atze Dijkstra, Universiteit Utrecht, working on EHC, interested to see what can be reused/learned/shared/etc.
Thomas DuBuisson, generally interested in GHC RTS
Edsko de Vries, Trinity College Dublin. Interested in using GHC as a backend (code generator) for an experimental functional language with a substructural (uniqueness) type system.
Kathleen Fisher, ATT, interested in adding support for data description to Haskell
Greg Fitzgerald, language enthusiast relatively new to Haskell (comment stolen from Mathew Mills :)
David Frech, interested in the "backend": STG code generation and its relationship with the RTS.
Jeremy Gibbons, University of Oxford, generally curious about GHC internals - especially advanced aspects of typing
Andy Gill, Galois, Previous contributor to GHC, wanting tools for high assurance Haskell.
Paul Graunke, Galois, interested in formal methods (Isabelle) and GHC intermediate representations
Thomas Hallgren, dedicated Haskell programmer, interesting in redoing hOp with GHC 6.6 and have the changes included in the main source repository.
Markus Lauer, want to understand better how GHC works
Brett Letner, Galois, interested in code coverage tool for Haskell.
Andres Löh, University of Bonn, interested in type systems and generic programming
Jake Luck, interested in interactive debugging and accessing GHC internals from embedded systems.
Ian Lynagh, interested in all parts.
John Matthews, Galois, interested in connecting GHC to the Isabelle theorem prover.
Luke Maurer, recent graduate of Carleton College, just looking to get my hands dirty
Mathew Mills, language enthusiast relatively new to Haskell.
Brandon Moore, Yahoo. Interested in types, syntax extension.
Ravi Nanavati, Bluespec. Interested in being able to fix GHC bugs before they burn me.
Echo Nolan, interested in optimization, absorbing intelligence by osmosis.
Andrew Pimlott, Planning Systems Inc., interested in the dark corners we stumble into during application development, and space profiling.
Norman Ramsey, interested in either code generation or refactoring some part of the run-time system (and would really like to learn more about how the run-time system is put together, even if it's too big for a weekend hack).
Doaitse Swierstra, Universiteit Utrecht, working on EHC, interested to see what can be reused/learned/shared/etc.
Aaron Tomb, University of California, Santa Cruz, interested in formal verification of Haskell programs, advanced type systems, and learning about the internals of GHC
Mark Tullsen, Galois, interested in tools for Haskell and high assurance run-time systems.
Malcolm Wallace, nhc98-hacker, wanting to steal ideas :-), particularly from the type-system.
Geoffrey Washburn, UPenn. Previous GHC contributor (original implementation of GADTs in Core). Wide variety of language interests involving expressiveness, security, and practical programming.
Peng Li, UPenn. Interested in multiprocessor GHC, STM, RTS and I/O libraries.
Kim-Ee Yeoh, interested in documenting, housekeeping, and general opening up of compiler sources to a wider audience
Jeffery Zhang, interested in type systems and metaprogramming.
Ahn, Ki Yung, PSU. GHC user. intersted in better data-sturcture libraries with more generic and powerful interface, eliminating dictionary passing cost of class functions anywhere possible, GADTs.
Lyle Kopnicky (lyle AT rentrak DOT com), interested in learning more about compiling Haskell.
Lee Pike, Galois. Interested in "Haskell on bare metal" and a lightweight/high-assurance RTS.
Troy Howard, Portland, discover-e Legal Inc., generally interested in compilers and oddball languages with a focus on iterative performance optimization.
Max Strini, Portland, discover-e Legal Inc., interested in type-level programming and optimization