This is the eighth edition of our GHC activities report where we describe the work on GHC and related projects that we are doing at Well-Typed. The current edition covers roughly the months of August and September 2021.
You can find the previous editions collected under the ghc-activities-report tag.
A bit of background: One aspect of our work at Well-Typed is to support GHC and the Haskell core infrastructure. Several companies, including IOHK, Facebook, and GitHub via the Haskell Foundation, are providing us with funding to do this work. We are also working with Hasura on better debugging tools. We are very grateful on behalf of the whole Haskell community for the support these companies provide.
If you are interested in also contributing funding to ensure we can continue or even scale up this kind of work, please get in touch.
Of course, GHC is a large community effort, and Well-Typed’s contributions are just a small part of this. This report does not aim to give an exhaustive picture of all GHC work that is ongoing, and there are many fantastic features currently being worked on that are omitted here simply because none of us are currently involved in them in any way. Furthermore, the aspects we do mention are still the work of many people. In many cases, we have just been helping with the last few steps of integration. We are immensely grateful to everyone contributing to GHC. Please keep doing so (or start)!
Currently, Ben Gamari, Andreas Klebinger, Matthew Pickering and Zubin Duggal are working primarily on GHC-related tasks. Sam Derbyshire has just joined the team at the start of October.
Many others within Well-Typed, including Adam Gundry, Alfredo Di Napoli, Alp Mestanogullari, Douglas Wilson and Oleg Grenrus, are contributing to GHC more occasionally.
Haskell Implementor’s Workshop
A few from our team presented various facets of their work at the Haskell Implementor’s Workshop in late August. More discussion of these presentations can be found in a previous HIW recap post on this blog.
Ben has been handling backports and release planning for the 9.2.1 and 9.0.2 releases.
Zubin fixed some bugs with LLVM version detection in the HEAD and 8.10.5 releases (#19973, #19828, #19959).
The bindists produced by hadrian now have a very similar structure to the ones produced by
make(Matt, !6345, !6349).
Compiler error messages
- Alfredo continued working on the conversion of GHC diagnostic messages from plain structured documents to richer Haskell types. After porting some errors in the driver code (!6249) he turned his attention to the modules in GHC’s typechecker (!6414), and he’s currently converting GHC’s typeclass-derivation code to use the new diagnostic infrastructure (!6561).
Matt has been fixing a number of bugs discovered after recent driver refactoring. Hopefully everything works as before in the 9.4 release! (!6508, !6507, !6412)
Matt attempted to implement the splice imports proposal but the specification didn’t correctly enforce level invariants. The latest idea is to introduce a complementary “quote” import to distinguish imports allowed to be used in quotations.
Haddock and documentation
- Zubin has been finishing up the the long-pending
hiHaddock work, which should allow Haddock to generate documentation using only GHC interface (
.hi) files (!6224). This greatly simplifies Haddock’s implementation, and allows it to skip parsing, renaming and type-checking files if the appropriate information already exists in the interface files, speeding it up greatly in such cases. This also reduces Haddock’s peak memory consumption. Identifiers in Haddock comments will also be renamed by GHC itself, and the results are also serialized into
.hifiles for tooling to make use of. A number of Haddock bugs were fixed along the way (#20034, haddock #30, haddock #665, haddock #921).
Profiling and debugging
- Andreas continued looking into using the machine stack register for the haskell stack. blog post. While we have a branch that uses the machine stack register there are issues with perf not unwinding properly as well as issues related to llvm compatibility. For these reasons we will likely stop looking into this for the time being.
Matt has been investigating several compiler performance regressions when compiling common packages (#19478, #19471).
Andreas improved further on tag inference in !5614. As it stands it improves compiler performance while also improving runtime for most programs slightly.
During investigation of some edges of the tag inference work Andreas found some edge cases in GHC’s current code generation (#20334, #20333, #20332).
Adam has been working on a new approach to improving compilation performance for programs with significant use of type families. The key idea is to introduce a more compact representation of coercions in Core, which will occupy less space and be faster to traverse during optimisation (#8095, !6476).
- Matt diagnosed and found an interesting recent regression in the text package benchmarks due to code generation using 8-bit instructions which causes partial register stalling (#20405).
Ben added build system support for multi-target native toolchains (e.g.
clang), allowing GHC to be used robustly on platforms which may run code for multiple platforms (#20162).
Ben fixed numerous linking issues affecting musl-based platforms, enabling static linkage on Alpine.
Ben fixed a number of build system bugs pertaining to libffi linkage affecting Darwin.
Ben fixed a Hadrian bug causing binary distribution installation to use platform parameters from the build environment instead of the installation environment (#20253).