Better Error handling

[TeXitoi]

The custom derive should dump great error message instead of just generating dumb code in case of bad usage.

[williamyaoh]

The way I'd want to do this would be to not use syn to parse the input tokens and instead have structopt do its own validation/parsing. syn is already making use of a nom parser, so the idea would be to rip out some code from syn and write a parser to generate data structures specifically for use by structopt. This would also make it easier to write the code generation itself; we wouldn't need asserts sprinkled around and such.

[TeXitoi]

I think this will be overkill to do that like that.

[CreepySkeleton]

@TeXitoi
Hi guys, I made a tool exactly for this purpose: https://github.com/CreepySkeleton/proc-macro-error . It should be relatively easy (mostly replace panics with span_error and similar) to integrate and I would do it if you don't mind. What do you think?

[TeXitoi]

I thought it was now integrated directly in syn, but it seems you know better than me the proc macro ecosystem ;-)

Sure, I'll accept with pleasure. For perfect errors, you might have to change some code around to not loose span information.

[CreepySkeleton]

For the record: AFAIK syn has nothing to do with error reporting, it's almost entirely about parsing. There is proc_macro::Diagnostics which is for error reporting but is still expiremental.

Just so you know - I am working on this, will make a PR in two days or so

[TeXitoi]

OK, seeing the span handling from a bit away in syn confused me about that.

I wait for your PR! Thanks!

[CreepySkeleton]

@TeXitoi By the way, I think we can close this

[TeXitoi]

Of course. Thanks again.