HyperQuark

HyperQuark compiles Scratch projects to WebAssembly, with the aim of running them quicker than scratch and also quicker than TurboWarp, which compiles to javascript. HyperQuark is not at feature parity with scratch, but is useable for a number of projects that only use the blocks listed on the home page. Projects listed on the homepage should function correctly, but are designed for debugging and so you may need to open up your browser's devtools (usually ctrl+shift+j/i) to see say/think block 'bubbles'.

Prerequisites

• Rust with a nightly toolchain - install via rustup
• the wasm32-unknown-unknown target (rustup target add wasm32-unknown-unknown)
• wasm-bindgen-cli (cargo install -f wasm-bindgen-cli)
• wasm-opt (install binaryen using whatever package manager you use)
• cargo-outdir (cargo install cargo-outdir)

Building

./build.sh -Wp # use -Wd for a debug build without optimisation

You may need to run chmod +x build.sh if it says it doesn't have permission.

The build script has additonal configuration options; run ./build.sh -h for info on these.

If you experience runtime stack overflow errors in debug mode, try using the -s or -z options to enable wasm-opt, or build in production mode; weird wasm errors in production mode may conversely be solved by disabling wasm-opt using the -o flag.

If an error occurs during execution and you need a stack trace, run with the -D flag to enable panicking with DWARF debug symbols. You need to set up your browser to display a proper stack trace using these debug symbols. You'll need to disable wasm-opt using -o for this to work, as wasm-opt currently crashes when it encounters DWARF.

To preview the website (e.g. to run a project), use npm run watch. Do not just run vite as this will build in debug mode which doesn't work.

Adding a new block

To add a new block named category_opcode, if it cannot be reduced to simpler blocks:

1. create src/instructions/category/opcode.rs. Make sure to use super::super::prelude::* and create the relevant pub items:

• (optional) pub struct Fields (must be Debug and Clone)
• pub fn wasm(func: &StepFunc, inputs: Rc<[IrType]>, (fields: &Fields)?) -> HQResult<Vec<InternalInstruction>>;
• • wasm is generated using the wasm_gen::wasm macro. See its README for usage instructions, or e.g. say.rs for an example.
• pub fn acceptable_inputs() -> HQResult<Rc<[IrType]>>;
• • these should really be base types (see BASE_TYPES in types.rs)
• pub fn output_type(inputs: Rc<[IrType]>, (fields: &Fields)?) -> HQResult<ReturnType>;
• • the output type should be as restrictive as possible; loose output types can cause us to lose out on some optimisations
• • Most output types should be either ReturnType::None or Singleton(IrType) (included in the module prelude); blocks can return multiple values via MultiValue(Rc<[IrType]>) but probably shouldn't.
• ensure to add relevant instructions_test!s - see instructions/tests.rs for usage

2. add pub mod opcode; to src/instructions/category.rs, creating the file if needed

• if you're creating the category file, add mod category; to src/instructions.rs

3. add the block to from_normal_block in src/ir/blocks.rs; in most cases this should be a direct mapping of BlockOpcode::category_opcode => IrOpcode::category_opcode
4. add the block's input names to input_names in src/ir/blocks.rs

If the block can be reduced to simpler steps, only carry out steps 3 and 4 above.

generared WASM module memory layout

name	number of bytes	optional?	description
stage info	8	no	src/wasm/mem_layout.rs
sprite info	80 * number of sprites (i.e. target num - 1)	yes	see src/wasm/mem_layout.rs
threads	4 * thread_num	present iff using the CallIndirect scheduler	indices of the next step funcs of currently running threads

Memory layout guarantees

All values should be guaranteed to have the maximum possible alignment for that value type. Unused/padding bytes are not guaranteed to have any specific values and may be overwritten.

License

Licensed under either of

• Apache License, Version 2.0 (LICENSE-APACHE or http://www.apache.org/licenses/LICENSE-2.0)
• MIT license (LICENSE-MIT or http://opensource.org/licenses/MIT)

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.