Philippe Laferrière
Philippe Laferrière
The current REPL executes everything typed by the user wrapped in a `begin ... end`. This means, amongst other things, that a user cannot define procedures and execute them. This...
As a follow-up to #1457. Context: - main issue comment: https://github.com/0xPolygonMiden/miden-vm/issues/1457#issue-2471210549 - PR discussion: https://github.com/0xPolygonMiden/miden-vm/pull/1581#discussion_r1853114792
As a follow-up to #1457, we should move some of the system events (formerly "advice injectors") introduced in #1581 to the standard library. Context: https://github.com/0xPolygonMiden/miden-vm/issues/1457#issuecomment-2491157516 and https://github.com/0xPolygonMiden/miden-vm/issues/1457#issuecomment-2491951354
The test [here](https://github.com/0xPolygonMiden/miden-vm/blob/b13ed4d67c13d99d750d7b91293a91b0185cb84d/miden/tests/integration/exec_iters.rs#L10).
> Does `Operation::Loop` allow for do/while-form loops rather than while/do loops? That would be super useful if so, but my understanding is we only have condition-guarded loops, which is a...
>instead of hardcoding this `divmod`, could we not use `u32divmod.8 = (q,r)` and treat `q` and `r` "dynamically"? We'd probably pay >Can we actually afford ~ 100 cycles x however...
In #1717, we defaulted to passing the `SourceManager` to the top-level `processor::execute`. We should investigate passing it in a different way - whether it's through the `MastForest`, `Host`, or something...
>Speaking as someone who has worked on a couple high-performance virtual machine implementations, the fastest technique I'm aware of for the core loop of a VM is computed goto, e.g....
Work towards #2406 Stacked on #2475. Migrates - all decoder tests to use `build_trace()` instead of `Process` - all operation-specific tests are migrated to test the `FastProcessor` instead of `Process`