Ideas and concepts that are not currently being pursued but may become relevant in the future. Nothing here is guaranteed to be implemented.
The compiler could detect data dependencies between function invocations (primarily through local scope and argument analysis) and schedule independent segments to run in parallel automatically.
Unless a function is explicitly declared sequential, the compiler would build a dependency graph and group independent segments into parallel batches, sized to amortise thread-synchronisation overhead (configurable).
Two possible execution models were considered:
- Batch execution — run a large group of independent segments until a synchronisation point, build the next batch, repeat.
- Dynamic execution — schedule segments as their dependencies are satisfied.
The long-term vision included compiling these parallel segments to run directly on GPUs or similar data-parallel hardware, using green threads on conventional CPUs.
A restricted type pairs a base type with a validation predicate (a substack that
receives the value and returns true if it is valid):
(0 gt) rint! # positive integers
The compiler can then prove at compile time whether a given literal satisfies the restriction, turning runtime failures into compile-time errors.