GNATCOLL.Opt_Parse has no concept of argument constraints

[LordAro]

That is, a "built in" way of specifying that if --foo is provided, --bar must also be given. Or --foo cannot be provided with --bar. And similar.

I'm happy to implement this myself and pull-request it, if such a thing is desirable. (Any thoughts on design and implementation of such a thing would be appreciated). I'd probably go with something similar to python's argparse groups - https://docs.python.org/3/library/argparse.html#argparse.ArgumentParser.add_mutually_exclusive_group

[raph-amiard]

Hi @LordAro,

I'm interested in the feature, albeit a bit wary, and certainly you have approximately 1800% more chances of getting this feature if you have a pull request for it 🙂

My only reservation is whether this is redundant with subcommands (that you also have in Python's argparse: https://docs.python.org/3/library/argparse.html#sub-commands). Most of the time, when I have mutually exclusive group of arguments, the whole argument parser would be better redesigned with subcommands. Would that be the case in your use case, and if not I would be curious to know a bit more about it.

But on a high level we would very much welcome new features and contributions to Opt_Parse, and having a way to express the above seems like a good idea.

Thanks for your time!