add coerce_empty_to_nil config for string conversion

[leifg]

After clarifying in #58 seems to be a good config flag for coercing empty strings into nil.

Closes #58.

[codecov]

Codecov Report

Merging #65 (12a93ac) into master (89e2d8d) will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #65      +/-   ##
==========================================
+ Coverage   99.22%   99.26%   +0.04%     
==========================================
  Files           6        6              
  Lines         386      409      +23     
==========================================
+ Hits          383      406      +23     
  Misses          3        3              

Impacted Files	Coverage Δ
lib/sorbet-coerce/converter.rb	97.50% <100.00%> (+0.13%)	⬆️
spec/coerce_spec.rb	100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us.

[mattxwang]

Thank you for the quick turnaround and your contribution @leifg! This seems pretty great to me, and I've had a chance to test this locally (as well as on CI - thanks for submitting the PR to resolve the issues with ruby 3+).

What are your thoughts on also adding a global configuration parameter, similar to TypeCoerce::Configuration.raise_coercion_error? No need to do this if you don't think it's necessary (for example, I'm not sure if it's relevant in your use-case of request parsing).

---

As an aside, I'll probably publish a new release of the gem after we merge this in. Hopefully you'll have a chance to test this soon!

[leifg]

What are your thoughts on also adding a global configuration parameter, similar to TypeCoerce::Configuration.raise_coercion_error? No need to do this if you don't think it's necessary (for example, I'm not sure if it's relevant in your use-case of request parsing).

For me personally I don't see a use case. In addition: making something like a global flag could cause unwanted errors because you would get instances of TypeError where an empty string is fine.

[mattxwang]

For me personally I don't see a use case. In addition: making something like a global flag could cause unwanted errors because you would get instances of TypeError where an empty string is fine.

Sounds good, just wanted to check in. Thanks again for your contribution, much appreciated!

[leifg]

Sounds good, just wanted to check in. Thanks again for your contribution, much appreciated!

Of course! Always happy to help out.