Add support for parsing bools,

[jonathanmorley]

this breaks the current behaviour of any attribute resolving to true

[RReverser]

Yeah, as you mentioned, this is a breaking change (it breaks even our use cases which I created this library for in the first place), and I don't think there's strong enough motivation to do it.

Unfortunately, there's too many ways to represent booleans in string form - some use 0/1, some absence/presence of attributes, some literals true/false (and then, they can be either case-sensitive or not), and there are many others. We will never be able to cover each case.

I think the existing way to represent bools is as good/bad as any other, and breaking compatibility here doesn't bring any obvious advantage, so it's better to leave it as is and have consumers with different needs make deserialization in their own way.

Also, you might want to track serde-rs/serde#745 which sounds like should fix this issue in a more generic way.