Remove generics from Error by making fragment a String

[RCasatta]

This cause a reduction of IR lines emitted by the compiler of about 2% (using cargo llvm-lines), so this should improve compile times.

I've seen using the parse example to check the produced binary size, but it's a small snippet I am not sure it calls a representative portion of the API surface.

From a library perspective this is equivalent since the field is used only to be converted to string.

In theory externally someone could use the field in the error to do things so it comes at a cost.

[apoelstra]

When no-std is on you need to explicitly import alloc::string::String I believe. This currently does not compile without std.

But concept ACK doing this. It seems like an improvement to me. Eventually we should go over all the error types like we have been doing with rust-bitcoin.

[tcharding]

One day I'd love to get time (ie months) to focus on miniscript.

[RCasatta]

no-std fixed

[apoelstra]

I fixed CI in #645. I'd like to get that in and rebase this because it's the kind of thing that we want CI to be passing for.

[RCasatta]

rebased to take #645

[apoelstra]

Looks like you need to run cargo fmt.

[RCasatta]

Looks like you need to run cargo fmt.

done

It's still broken on nightly which seems to have alloc::string::String automatically imported, I am not sure how to fix

[apoelstra]

When std is on then String, ToString etc are always imported. It's just that nightly warns about this. There are half a dozen open issues on rustc related to this new lint triggering all the time in counterproductive ways.

Meanwhile to deal with it you just need to make sure your imports are feature-gated on nostd.

[RCasatta]

ah right the same you did here

[apoelstra]

ACK 0afd832

[apoelstra]

cc @sanket1729 can you concept-ACK this?

[apoelstra]

On further thought I'm just going to merge this. I have some bigger ideas about how to handle errors in this crate, and getting rid of these generics will make future refactoring much easier.