skrobinson

I oppose merging this change at this time. I'd prefer to leave this PR open so any others with this error have a possible fix and can upvote your solution....

My thinking is that the build errors you and @marstaik are seeing are `clang-cl` bugs, not `argparse` bugs. Have you tried submitting an Issue with clang-cl? But maybe I'm wrong...

`argparse` is a C++17 project and GCC 7 does not fully support C++17. My preference is to not add complexity to extend support back to older compilers. You could use...

No, `argparse` will not ignore unrecognized arguments since v1.6. I recommend leaving this issue open. Someone may see this and decide to add [parse_known_args](https://docs.python.org/3/library/argparse.html#argparse.ArgumentParser.parse_known_args).

Unlike the Python `argparser`, this `argparse` does not return a `Namespace` object from `::parse_args`, but stores values in the `ArgumentParser` instance. This storage is not assignable except through `::parse_args`. A...

Thank you for your report. Let me look into this and get back to you next week.

Digging into the project history, I'm guessing the expected practice is to disable the parent parser default arguments. But that paragraph did not get copied into the README. You can...

I see your point and agree that it can be useful to run an action function with the `.default_value`. But I do not see a simple way to add this...

I'm not a fan of external underscores for two reasons: (1) more easily visually missed; (2) I've seen an editor that chopped off the bottoms of rows of text with...

> I'm not particularly bothered by the mixed use of class and struct - the naming style was meant for both classes and structs. In your experience, is there a...