WASM Improvements

[Marcel-Nordeck]

This brings a few issues I noticed while trying to port https://gitlab.opencode.de/fitko/matrix-g2x/fit-neo-web/-/merge_requests/3 from the filament Wasm mega PR over to the main branch of the SDK. Without these (and downgrading uniffi 0.29.4 due to jhugman/uniffi-bindgen-react-native#316 ) the bindings cannot compile.

Note that also the inline stubs seem to be required as uniffi does checksums over the actual function signature including any attributes. This seems to happen before evaluation of them, however. Which caused checksum mismatches if I used nicer looking cfg attributes (splitting Wasm and non Wasm world into 2 functions instead of inline blocks). Ideally, one day uniffi gets smarter about that, but it doesn't seem to be a thing yet. https://mozilla.github.io/uniffi-rs/latest/proc_macro/index.html#conditional-compilation seems to also not help us here.

One more thing I noticed is that while the indexeddb support was added, the bundled-sqlite feature still stayed as a default feature. This causes issues on Wasm as SQLite can't be compiled on that platform using the sqlite-sys package which rust-sdk utilizes. (It will complain about missing headers). Therefore, I think it's easiest to just remove it from the defaults and have it active conditionally by the consumer.

Finally, WASM threading support currently lacks Send, so uniffi has to be told to be less strict about their requirements to support this. This might change in the future, however, if Wasm gets proper threading support in rust land. Additionally, there was some code using futures-executor, but the dependency was not added to the wasm platform.

---

• Public API changes documented in changelogs (optional)

---

(This PR is made with my Nordeck 🎩 for the FITKO Matrix-G2X project over at https://gitlab.opencode.de/fitko/matrix-g2x/gitlab-profile )

[Marcel-Nordeck]

Hm not entirely sure why it fails on the feature flag for uniffi as the workspace Cargo.toml claims 0.30.0. I know it works with 0.29.4 upwards. I am guessing it may just need a new lockfile right?

[Marcel-Nordeck]

Also sorry for the force pushing. Realised the conventional commit policy too late. From now on, I will instead use fixup commits when needed.

[poljar]

Hm not entirely sure why it fails on the feature flag for uniffi as the workspace Cargo.toml claims 0.30.0. I know it works with 0.29.4 upwards. I am guessing it may just need a new lockfile right?

complement-crypto, which is used for one of the CI runs, downgrades to 0.28.x since uniffi-bindgen-go doesn't support 0.29 or 0.30 yet.

I think only enabling this feature under Wasm, which makes sense regardless of complement-crypto, would be one way to solve this problem.

[codspeed-hq]

CodSpeed Performance Report

Merging #5838 will not alter performance

_{Comparing Marcel-Nordeck:MTRNord/wasm-improvements (6a1c74c) with main (fa6d18b)}

Summary

✅ 50 untouched

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.60%. Comparing base (fa6d18b) to head (fd93a0f).
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #5838   +/-   ##
=======================================
  Coverage   88.60%   88.60%           
=======================================
  Files         361      361           
  Lines      102150   102150           
  Branches   102150   102150           
=======================================
+ Hits        90508    90509    +1     
+ Misses       7425     7423    -2     
- Partials     4217     4218    +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[poljar]

Hm not entirely sure why it fails on the feature flag for uniffi as the workspace Cargo.toml claims 0.30.0. I know it works with 0.29.4 upwards. I am guessing it may just need a new lockfile right?

complement-crypto, which is used for one of the CI runs, downgrades to 0.28.x since uniffi-bindgen-go doesn't support 0.29 or 0.30 yet.

I think only enabling this feature under Wasm, which makes sense regardless of complement-crypto, would be one way to solve this problem.

Well ok, I updated complement-crypto here matrix-org/complement-crypto#215.