fix forloop codegen over non-split word arrays of length >= 64 elements

This commit is contained in:
Irmen de Jong
2025-05-13 22:43:20 +02:00
parent e31ef6f06f
commit 67231af623
4 changed files with 22 additions and 79 deletions

View File

@@ -60,6 +60,7 @@ IR/VM
Libraries
---------
- Add split-word array sorting routines to sorting module?
- Add double-array sorting routines to sorting module? (that allows you to sort a second array in sync with the array of numbers)
- cx16: _irq_dispatcher now only dispatches a single irq source, better to ROL/BCC to handle *all* possible (multiple) sources.
- See if the raster interrupt handler on the C64 can be tweaked to be a more stable raster irq
- pet32 target: make syslib more complete (missing kernal routines)?
@@ -70,7 +71,6 @@ Libraries
Optimizations
-------------
- Sorting module gnomesort_uw could be optimized more by fully rewriting it in asm? Shellshort seems consistently faster even if most of the words are already sorted.
- Compare output of some Oscar64 samples to what prog8 does for the equivalent code (see https://github.com/drmortalwombat/OscarTutorials/tree/main and https://github.com/drmortalwombat/oscar64/tree/main/samples)
- Optimize the IfExpression code generation to be more like regular if-else code. (both 6502 and IR) search for "TODO don't store condition as expression"
- VariableAllocator: can we think of a smarter strategy for allocating variables into zeropage, rather than first-come-first-served?