wasm-bindgen: Added support for Internet Explorer 11

[zer0x64]

It is a pretty obscure use case, but I did stumble upon it:

When using wasm2js to compile WASM to Javascript for use in older browsers as recommended by the wasm-bindgen documentation, we found out that rand crate did not work on Internet Explorer 11 since the window.crypto object is named window.msCrypto.

This PR adds a check to use msCrypto if crypto is not present.

[josephlr]

Thank you @zer0x64 for this patch. If Rust docs reccomend wasm2js, and supporting IE isn't that hard, we should do it. Note, we would probably not support IE versions before 11.

Quick question, is there an easy way to test this in CI? I don't think so, but I thought I would ask.

[josephlr]

Formatting nit here:

typeSelf_;#[wasm_bindgen(method, getter, structural)]fncrypto(me:&Self_) -> JsValue;#[wasm_bindgen(method, getter, js_name = "msCrypto", structural)]fnms_crypto(me:&Self_) -> JsValue;

We should explain the use of msCrypto above (when we use it)

[josephlr]

We should document the crypto vs mscrypto distinction here, so it's all in one place. Something like

// If `self` is defined then we're in a browser somehow (main window// or web worker). We get `self.crypto` (called `msCrypto` on IE), so we// can call `crypto.getRandomValues`. If `crypto` isn't defined, we// assume we're in an older web browser and the OS RNG isn't available.

[josephlr]

We can make this logic more straightforward using Rust's match guards:

let crypto:BrowserCrypto = match(self_.crypto(), self_.ms_crypto()){(crypto, _)if !crypto.is_undefined() => crypto.into(),(_, crypto)if !crypto.is_undefined() => crypto.into(),
_ => returnErr(BINDGEN_CRYPTO_UNDEF),};// Test if `crypto.getRandomValues` is undefined as wellif crypto.get_random_values_fn().is_undefined(){returnErr(BINDGEN_GRV_UNDEF);}

[zer0x64]

I really like that!

[josephlr]

@zer0x64, also be sure to run cargo fmt --all -- **/*.rs before submitting. By default, cargo fmt won't format the target-specific files.

[zer0x64]

Great, I'll address the changes tomorrow!

For the CI, unfortunately, wasm-bindgen-test doesn't support headless IE, so the only way I see to test this in a CI is to use a framework like Selenium. It is certainly doable, but not as straightforward as testing Firefox/Chrome. Also, I don't think this can be done without a Windows build server. However, I can help integrate it if you decide to add this check to the CI!

EDIT: Also forgot to mention that since IE does not support webassembly, we would need to use wasm2js to translate it to javascript and, from experience, that part is mostly manual work and does not integrates well in a CI for now. According to the wasm-bindgen doc, some day it should be supported out of the box without manual work.

[josephlr]

@zer0x64 would wasm-bindgen's vendor_prefix do all of this for us?

Could we just do:

 type Self_;
#[wasm_bindgen(method, getter, vendor_prefix = "ms", structural)]
fn crypto(me: &Self_) -> JsValue;

and not need any additional code? I don't know if msCrypto and mscrypto are the same thing in JS.

EDIT: Also forgot to mention that since IE does not support webassembly, we would need to use wasm2js to translate it to javascript and, from experience, that part is mostly manual work and does not integrates well in a CI for now. According to the wasm-bindgen doc, some day it should be supported out of the box without manual work.

This seems reasonable. We can always add more testing later once the tools are in a better state. Our coverage right now is fairly good (Node, Chrome, Firefox).

[zer0x64]

I did some tests with the vendor_prefix and it simply doesn't seem to work in this context at all. I think it might be since the annotation is made for type and not for fn. Although even if it worked, we'd still have the problem that it exports msCrypto and not mscrypto.

[zer0x64]

I've just pushed the fixes for your comments!

[josephlr]

We just need #141 to be approved, then we can merge this.

[josephlr]

@zer0x64 can you rebase this onto the 0.2 branch (right now this PR is against the master branch), this will allow you to pickup #141

The only change you'll need will be to turn BINDGEN_CRYPTO_UNDEF into Error::BINDGEN_CRYPTO_UNDEF

I would do this myself, but I don't have access to modify your PR.

EDIT: Actually, having this PR against the master branch is fine.

[josephlr]

Actually, @zer0x64, having this against the master branch is fine. I'll handle putting the changes on the 0.2 branch. Just rebase this on #142 when that lands.

[josephlr]

@zer0x64#142 has been merged, can you rebase this onto master?

[josephlr]

We no longer need this line (as we already convert the JsValue to a BrowserCrypto above.

[zer0x64]

No problem!

[zer0x64]

Rebased and tested on my side, so when the CI pass it will be ready to merge!