Conflicts don't propagate from the global level to the subcommand

[larsrh]

As discussed on Gitter: https://gitter.im/kbknapp/clap-rs?at=5a9f003af3f6d24c68450335

Affected Version of clap

2.31.1

Expected Behavior Summary

When declaring that a subcommand argument is in conflict with a global argument, clap should produce an error no matter whether the global argument comes after or before the subcommand.

Actual Behavior Summary

global argument comes after: conflict
global argument comes before: no conflict

Sample Code or Link to Sample Code

(by @kbknapp)

extern crate clap;
use clap::{Arg, App, SubCommand};

fn main(){
    let _ = App::new("fake")
        .arg(Arg::with_name("glob")
            .short("g")
            .global(true))
        .subcommand(SubCommand::with_name("foo")
            .arg(Arg::with_name("conf")
                .short("c")
                .conflicts_with("glob")))
        .get_matches_from(vec!["fake", "foo", "-g", "-c"]);
}

Steps to Reproduce the issue

Run the above code with fake foo -g -c and fake -g foo -c and observe the difference.

[kbknapp]

Thanks for submitting!

For my own notes and from the gitter chat

• The flag is actually propagated down to the subcommand correctly, but the conflict doesn't register
• The same happens if we use options instead of flags (so it's not flag specific)

[mainrs]

Just a side note, shouldn't the help also not display the global arg if the subcommand arg is in conflict and the subcommand arg is required? This case wouldn't allow the global arg anyway to be passed down.

extern crate clap;
use clap::{Arg, App, SubCommand};

fn main() {
	let _ = App::new("fake")
		.arg(Arg::with_name("glob")
			.short("g")
			.global(true))
		.subcommand(SubCommand::with_name("foo")
			.arg(Arg::with_name("conf")
				.short("c")
				.takes_value(true)
				.required(true)
				.conflicts_with("glob")))
		.get_matches_from(vec!["fake", "foo", "-g", "-c"]);
}