Fix std::string_view support in transforming validators

[peterazmanov]

Transforming validators convert const char * to std::string, but don't do the same for std::string_view. This causes UB in details::lexical_cast that returns std::string_view holding dangling pointer.

[codecov]

Codecov Report

Merging #300 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master   #300   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files          12     12           
  Lines        2880   2880           
=====================================
  Hits         2880   2880

Impacted Files	Coverage Δ
include/CLI/TypeTools.hpp	100% <ø> (ø)	⬆️
include/CLI/Validators.hpp	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e6aca64...81e5393. Read the comment docs.

[codecov]

Codecov Report

Merging #300 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master   #300   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files          12     12           
  Lines        2880   2886    +6     
=====================================
+ Hits         2880   2886    +6

Impacted Files	Coverage Δ
include/CLI/TypeTools.hpp	100% <ø> (ø)	⬆️
include/CLI/Validators.hpp	100% <100%> (ø)	⬆️
include/CLI/App.hpp	100% <0%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e6aca64...f2fa90a. Read the comment docs.

[henryiii]

I'm thinking of moving the string_view part to the test, and using CLI11 + StringView would require a user add the overload. I'm not fond of adding a new include, it adds a dependency that can sometimes fail (as seen in the tests). I was hoping I could add a declaration instead of an include, but I'm not sure how to do that since it's really std::__1::basic_string_view and I don't think I want to rely on the __1 being there.

The other change is fine, adding a test is fine, and it might be something improved by #285.

[peterazmanov]

I'm not fond of adding a new include, it adds a dependency that can sometimes fail (as seen in the tests).

std::string actually includes <string_view> for __cplusplus >= 201703L. So there is no additional overhead.

... and using CLI11 + StringView would require a user add the overload.

I don't know how to prevent usage of std::string_view without adding the overload.

I was hoping I could add a declaration instead of an include, but I'm not sure how to do that since it's really std::__1::basic_string_view and I don't think I want to rely on the __1 being there.

AFAIK adding forward declarations for STL classes is not allowed by the standard.

[henryiii]

std::string actually includes <string_view> for __cplusplus >= 201703L. So there is no additional overhead.

I don't think that's required, but it's good enough for me.

AFAIK adding forward declarations for STL classes is not allowed by the standard.

Yes, I believe so, which is why they are fine to put them in inline namespaces...

it adds a dependency that can sometimes fail (as seen in the tests)

Okay, this was my mistake, now it looks fine.