Rewrite the WebCrypto ECDH using wasm-bindgen #980
Conversation
ghost
assigned 
|
Note that while it is not great to assume that we are in a browser environment when compiling for WASM, it is intended that we eventually merge |
|
|
||
| /// We use a `SendWrapper` from the `send_wrapper` crate around our JS data type. JavaScript data | ||
| /// types are not `Send`/`Sync`, but since WASM is single-threaded we know that we're only ever | ||
| /// going to access them from the same thread. |
There was a problem hiding this comment.
Unless the Threads feature gets implemented eventually. I think Chrome versions >= 70 already support it.
There was a problem hiding this comment.
Well, I don't know how to solve that and suggest opening an issue instead of blocking this PR. It is not a safety problem, only a panicking problem.
| let _ = tx.take() | ||
| .expect("JavaScript promise has been resolved twice") // TODO: prove | ||
| .send(out); | ||
| // Note: contrary to what one might think, we shouldn't add the "deriveBits" usage. |
There was a problem hiding this comment.
Could you explain that bit?
There was a problem hiding this comment.
The importKey function takes a list of "usages": https://developer.mozilla.org/en-US/docs/Web/API/SubtleCrypto/importKey
For me it was obvious that one should pass the "deriveBits" usage, but if I remember correctly in practice both Firefox and Chrome produce an error if I do that.
It was untested and not working anyway. We don't lose anything by rewriting it.
This PR is tested and works against the Rust secio implementation.
cc #23