0. Overview (SPJ)

• The overall organisation of GHC
• Source Tree Roadmap
• the compilation pipeline

1. Building (SM)

• Getting the code (do a live build...)
• Setting up the build
• Building
• tweaking & recompiling, etc.
• how libraries/packages are built
• ghc-inplace vs. installed ghc, how does it run inplace (find its bits)
• how to "use" the build system and common tasks, pointer to docs
• using/extending the testsuite

2. The Compiler

• SPJ inside-out view: start with a single module (HscMain? downwards)
• SPJ the main datatypes (RdrName?, Name, Id, Var, TyVar?, Type, Core, HsSyn?, Cmm).
• SPJ renamer, typechecker, desugarer, core->core
• SPJ ModIface?, ModDetails?, ModGuts?
• SPJ Core->CorePrep?->Stg->Cmm
• SM PrimOPs: primops.txt.pp - what is generated from it?
• SM GHC API
• SM HscMain? upwards: ModSummary?, Finder, upsweep, downsweep,

3. Back end (SM)

• SM execution model STG + eval/apply
• SM Cmm->NCG
• SM Cmm->BCO (simple compilation scheme, no primops: GHC.PrimopWrappers?)
• SM Cmm->C

4. The Runtime (SM)

• .c vs. .cmm files
• diff between .cmm and C--
• roadmap
• heap objects, info tables
• architecture: storage manager, MBlock, Blocks, Storage, GC
• architecture: scheduler, capabilities, tasks
• FFI (adjustors)
• Linker
• BCO interpreter
• asyncrhonous & synchronous exceptions
• STM