Dave Rolsky

For anyone else who finds this, here's the `Cargo.toml` line to get the working version: ``` grpc-web-client = { version = "~0.1", git = "https://github.com/titanous/grpc-web-client", rev = "dd15cde2f36ab10546c332a7bbcb55b1738018a5" } ```

I just encountered this via testify as well. It would be great to apply the fix in #111 or some other fix.

It'd probably be good to also get rid of all the `unwrap` calls in `call.rs` too. I can make a separate PR for that following the same pattern as this.

I reported this to GitHub as well, since what they're doing is simply wrong. https://github.com/orgs/community/discussions/78614

> It wouldn't solve the issue of non-graphql-compliant bodies though, we would have to have an extra variant somewhere for bodies that aren't valid graphql responses. I think the right...

> But I don't think our strftime-format string is going to be flexible enough to provide optimal localized formatting. And I don't think we have the experience to come up...

> Upon further testing by dumping out some mods from the ICU data I see results like: > > ```rust > #[allow(non_snake_case,non_camel_case_types,dead_code,unused_imports)] > pub mod aa_DJ { > pub const...

One more thing re: CLDR formatting. If you want to support this I think it's best to come up with new functions that accept these formats, like `format_cldr` or something...

Dupe of #22 and several others, I think. Basically the docs lie completely about the actual behavior on Windows. It _always_ invokes a shell there, regardless of what sub you...

Yeah, I agree. I just encountered this issue myself.