amethyst / amethyst/rustrogueliketutorial
Sharing save/load tricks (DRY, WASM)
- Vorherrschende Sprache
- Rust
- Sterne
- 970
- Forks
- 166
- PR-Merge-Kennzahlen
- Keine gemergten PRs in 30 T.
Beschreibung
Hey @thebracket! Thank you for the amazing library and book, I'm having a blast following along. This is not so much an "issue", but rather sharing what I learned in case you want to use it in the book at some point. I'll do it tl;dr style to respect your time:
* I've added saving / loading under WASM in https://github.com/abesto/rktrl/commit/b0da9c00125ba9bc2335f0403576b28e56c063f5. Basically it uses local storage (base64 encoded, because I GZip the saved file on the cross-platform code path).
* FWIW, the sentence "Supporting saving via LocalStorage (a browser/JavaScript feature) is planned for a future version of RLTK." confused me for a second - RLTK doesn't explicitly support saving / loading outside the browser either, it's all `serde` and `specs` (AFAICT at least).
* Reaching for increasingly big hammers, I managed to set up a procedural macro so that for the save / load systems there is zero repetition of the components (check out [the macro](https://github.com/abesto/rktrl/blob/b0da9c00125ba9bc2335f0403576b28e56c063f5/rktrl_macros/src/lib.rs) and [its usage](https://github.com/abesto/rktrl/blob/b0da9c00125ba9bc2335f0403576b28e56c063f5/src/systems/saveload.rs#L34-L60)). The big idea here is using the macro to split the tuple of components into chunks of at most 16, so that we can then rely on the the `specs_derive`-provided implementations for those.
* I'm toying with the idea of creating a similar macro to cut down on the boilerplate involved with creating system data structs (but that's less relevant as currently the tutorial doesn't use them)
* Getting things to compile for WASM in the first place was a bit tricky as dependencies pulled in parallel Specs features in their `default-features`, which trigger threading. Had to disable them in `cargo.toml` (see https://github.com/abesto/rktrl/commit/72a7a119385ca6ae5c6871dd31d281ba842d0ffe). I guess there might be a better way of doing this (maybe controlling whether a parallel dispatcher is used from code).
Finally, I wanted to add that I know how overwhelming input about big huge changes to writing projects can be, so wanted to explicitly say: no pressure, take these when and if they're useful. Even until then, maybe this will in itself be useful for someone else. Thank you again!
Beitragsleitfaden
Für dieses Repository ist kein Beitragsleitfaden indexiert
Bewertung
Dieses Issue wurde noch nicht bewertet.