GlobSet: the order of patterns added to the builder matters and it shouldn't

[calixteman]

With globset 0.4.2, the following test case is failing:

extern crate globset;
use globset::{Glob, GlobSetBuilder};

fn main() {
    let mut glob_builder = GlobSetBuilder::new();	
    for i in vec!["libcore/*", "libstd/*"] {
        glob_builder.add(Glob::new(&i).unwrap());
    }
    let globset = glob_builder.build().unwrap();
	
    assert!(globset.is_match("libstd/io/buffered.rs"));
    assert!(globset.is_match("libcore/char/methods.rs"));

    let mut glob_builder = GlobSetBuilder::new();	
    for i in vec!["libstd/*", "libcore/*"] {
        glob_builder.add(Glob::new(&i).unwrap());
    }
    let globset = glob_builder.build().unwrap();
	
    assert!(globset.is_match("libstd/io/buffered.rs"));
    assert!(globset.is_match("libcore/char/methods.rs"));
}

with (this is the last assertion in the code):
'assertion failed: globset.is_match("libcore/char/methods.rs")', src/main.rs:22:5

[calixteman]

@BurntSushi, any chance to have a fix for this bug ?

[BurntSushi]

@calixteman Thanks for the ping, but I haven't even had time to confirm this and I don't know when I will be able to look at fixing it.

[marco-c]

@BurntSushi could you release a new version of globset with the fix for this?