Accept also None or Objects as params

[RCasatta]

according to the jsonrpc spec params can be missing, or an object too.

close #107

Breaking change

[apoelstra]

concept ACK. But looks like certain feature combos are broken (see CI)

[RCasatta]

I thought after the first approval the other times the CI would start automatically, but it's not the case, I hope to have "blindly" fixed it.

[RCasatta]

I think to have tested all the cases launched with the CI locally, sorry for the back and forth

[apoelstra]

I thought after the first approval the other times the CI would start automatically, but it's not the case, I hope to have "blindly" fixed it.

All good. I blame Github, not you :).

Hit CI. Let's see how it goes.

[apoelstra]

@RCasatta can you add a unit test that demonstrates how to provide a list, and maybe a CHANGELOG entry explaining this? This will be a breaking change which will be particularly annoying because the RawValue API is so annoying.

[RCasatta]

I had to force push because there were a couple of reference of &[] not changed and not tested in CI (the one in the Readme and the one in integration_test/src/main.rs)

Moreover, I added:

• the unit test commit
• a call getblockhash in the integration test with both list and named configuration (not launched in CI though)
• update the changelog

[apoelstra]

Looks like some clippy lints are causing CI to fail.

[RCasatta]

Sorry again, should be fixed

[apoelstra]

Previously you had to provide a list with each thing wrapped in arg(), which was bad enough ... but now it looks like you need to wrap your lists in Some(to_raw_value(&json!(actual_thing)).unwrap()) which is way worse.

I think we need a better API here.

[RCasatta]

You don't need the json! call when you have a struct with serde::Serialize

[apoelstra]

@RCasatta ah, I gotcha. So in your list unit test, where you write json!([0]) it's actually fine to just do to_raw_value(&[0]) (I tested this and it worked).

Ok, I'm happy then.