diff options
| -rw-r--r-- | Notes/todo.txt | 57 |
1 files changed, 43 insertions, 14 deletions
diff --git a/Notes/todo.txt b/Notes/todo.txt index 3ea2f69d..5b244a34 100644 --- a/Notes/todo.txt +++ b/Notes/todo.txt @@ -6,33 +6,62 @@ TODO: - Remove variables that are just assigned from arguments - Clean up AST - Almost done, just a few little niggles left -- Split arg patterns and arg types up for cleaner serialisation - - Could be a good idea to do as part of changing HIR::TraitImpl to only contain the impl data - - May not be too useful due to argument monomorphisation. - Optimise typecheck. + - Mostly done with trait resolve cleanup ## Big change TODOs - Support MIR-only RLibs + - Defer C codegen until final binary generation? +- Dylib support (requires trans API restructure) + - Will improve disk usage for tests (can dynamically link libstd) - Fix Span annotations + - Spans are missing on HIR items, sub-par reporting in typecheck and later + - Spans can be off by a line or two - Refactor parse to use a consume model lexer - Optimise optimise (and typecheck) + - Partially down with trait resolution optimisation. Needs further profiling - Complete structed C codegen -- Switch MIR to be SSA (or have a SSA form) -- MIR restructure specified in MIR-PackedLValue.txt - - Designed to remove pointer chasing when doing MIR analyis (and save space) + - Upside: It'll look cool + - Unknown: Will it be faster? +- RTL/register-based SSA interrim backend + - Convert MIR into a form that can be handed over to LLVM or Cranelift (or GIMPLE) + - Use alloca-s for non-pointer/integer/borrowed locals ## Smaller changes -- Only generate destructors if needed (removes C warnings) +- Make type ascritpion its own node type + - AST and HIR + - Then remove per-node type annotations from AST - Cache specialisation tree -- Dependency files from mrustc - - Partally done, not complete + - TODO: Profile to determine if this is a slow point. - Delete HIR after MIR generation -- Use a more space-efficient (and readable) MIR LValue debug output - - e.g. `_0**.1#2` for Downcast(2, Field(1, Deref(Deref(Local(0))))) - - Will simplify reading of debug, and reduce the size of required debug - output - - Trailing dereference might be confusing with multiply operator? + - Problem: HIR is sometimes touched again after MIR gen + - May just be able to do the delete in the main MIR gen phase +- Split types and patterns in HIR function arguments + - Upsides: + - Less data in the serialsed .hir file + - Note: Patterns aren't actually stored in metadata + - Simpler logic post MIR generation + - Reduced use of .first/.second + - Memory usage reduction for external functions + - Downsides: + - Lots of code touched + - Extra complexity in typecheck and MIR lowering? +- Sort trait impls in a similar way to type impls + - ```c++ + template<T> struct ImplGroup { + std::map<SimplePath, T> named; + std::vector<T> primitives; + std::vector<T> generic; + } + std::map<SimplePath, ImplGroup<TraitImpl>> trait_impls; + ``` <!-- `--> + - Problem: Trait lookup on ivars? + - TODO: Profile a large crate again to see what the overhead of searching + trait impls is. + - Should help with widely implemented traits (e.g. Debug) + - Downside: Another SimplePath per impl cluster, plus the vector size - Not + too big but something to be aware of (7 pointers overhead plus indirection). ## Optimisations |