core: Inferencer should treat item assignments differently than direct assignments #482

New Issue

lyken · 2024-07-31T10:44:23+08:00

lyken commented

2024-07-31 10:44:23 +08:00

The implementation of the type inferencer should consider item assignments differently, i.e., target[key] = value. Currently, the current implementation is mixed with direct assignments, e.g., my_var = value.

The Inferencer should have a fold_setitem(target, key, value) and the CodeGenerator should have a gen_setitem(target, key, value) to handle target[key] = value with better control.

The implementation of the type inferencer should consider item assignments differently, i.e., `target[key] = value`. Currently, the current implementation is mixed with direct assignments, e.g., `my_var = value`. The `Inferencer` should have a `fold_setitem(target, key, value)` and the `CodeGenerator` should have a `gen_setitem(target, key, value)` to handle `target[key] = value` with better control.

lyken added the

low-priority

label 2024-07-31 10:45:27 +08:00

lyken referenced this issue

2024-07-31 10:51:01 +08:00

core: Inferencer `fold_stmt()` on `StmtKind::Assign` should not force unify `value_ty` with `target_ty` for `__setitem__` assignments. #460

lyken referenced this issue from a commit

2024-07-31 10:58:00 +08:00

WIP: core/ndstrides: implement ndarray subscript assignment and stuck on #482

lyken changed title from ~~core: Inferencer should treat `__setitem__` assignments differently than direct assignments~~ to core: Inferencer should treat item assignments differently than direct assignments

2024-07-31 14:37:01 +08:00

lyken referenced this issue

2024-08-02 15:31:11 +08:00

core: reimplement assignment type inference and codegen + other minor changes & refactor #483

lyken removed the

low-priority

label 2024-08-02 15:45:41 +08:00

lyken commented

2024-08-06 09:47:27 +08:00

Fixed by #483. Closing.

Fixed by https://git.m-labs.hk/M-Labs/nac3/pulls/483. Closing.

lyken closed this issue