Allow checking for any command exit value.

[vincentdephily]

This is pretty much what Assert::fails_with(self, expect_exit_code: i32) provides, except that an exit code of 0 is accepted. The current implementation fails the assertion if expect_exit_code=0 and the command returns with an exit code of 0.

Rationale for the functionality: I'm testing for specific exit codes with something that looks like:

        let t = vec![
            (&["all","fine"], 0),
            (&["user","error"], 1),
            (&["program","error"], 2),
        ];
        for (args, exit) in t {
            Assert::main_binary().with_args(args).fails_with(exit).unwrap();
        }

But this fails unless I special-case exit==0 to remove the fails_with() call. It's not hard, but it's ugly and surprising.

I'm not sure if fails_with() should be modified to accept 0 as an argument (ie this issue is a bug) or if a new function (say exits_with()) should be created (ie this issue is a feature request). Tell me which and I'll try to send a PR.

assert_cli 0.5.4 rust 1.25.0

[epage]

Yeah, the problem with fails being in the name, it implies more than just an exit code check.

The good news is that in the API refactor that is under way, we are narrowing the focus of the call to just be about asserting the exit code

See https://github.com/assert-rs/assert_cli/pull/98/files#diff-518bd9db9fa60ba5991cb1df6489cb2fR509

[vincentdephily]

Ok, so no need for a PR on this isolated issue and we can just close it when pr #98 lands ? Or will the new API take a while to arrive and an interim fix is useful ?

[epage]

If you need it now and want to create a PR, go ahead. #98 is still early in the process.

[epage]

Addressed in assert_cmd
https://github.com/assert-rs/assert_cmd
assert_cli is going to morph into a wider-focused tool, including things like assert_fs. See #41