Parse at least one argument

[lambda-fairy]

I'm writing a program that takes multiple arguments and performs an operation on each. Unfortunately, the arguments parser succeeds even when I supply no arguments, even though it doesn't make sense for my application.

Is there a way to force arguments to parse at least one argument?

[dag]

You want something like, some argument. Have a look at Alternative which the parser is an instance of.

[lambda-fairy]

Thanks – I've tried using some (argument ...) and (:) <$> argument ... <*> arguments ... and they both share the same problem. When I invoke the program with the --help option, the argument parser gobbles it up, and then my program tries to process it rather than displaying help as it should.

[dag]

Ah yea, that issue was fixed in arguments but not argument. I would like the former to be simply something like many argument but perhaps that's non-trivial or not possible for some reason? In the mean time you should be able to work around it by having the reader return Nothing if the argument starts with a hyphen.

[lambda-fairy]

That works – thank you again.

The inconsistent behavior should probably go in a separate issue, so I'm closing this for now.

[pcapriotti]

The problem here is that arguments has some special logic to:

• initially ignore arguments starting with '-'
• accept '--'
• accept arguments starting with '-' after '--' is encountered

Given this desired behavior, arguments cannot simply be implemented as many argument.

What we can do to make common use cases like the one in this Issue easier to
deal with, is to add a bunch of convenience builders, like:

• arguments1, non-empty argument list, with the same behavior as arguments
• argument', parse 1 argument, ignoring things starting with '-'

This way, many argument' would be similar to arguments (without the special
handling of --), and some argument' to arguments1.

Suggestions for better names are welcome. :)

[pcapriotti]

Fixed by 700e07b,

[dag]

I'm sort of pondering if this should happen at a lower level, splitting the input before parsing?