[enhancement] Support multiple test cases for a single test proc

[rw1nkler]

What's hard to do? (limit 100 words)

When using test procs, it is not possible to run different test cases, that will be eventually listed as separate entries in the report without duplicating test procs. It would be helpful to describe multiple tests cases that can reuse the same config() function.

Current best alternative workaround (limit 100 words)

Currently, the user has to either put all the tests in the single next() function, making individual test cases indistinguishable or create multiple test procs, which leads to significantly longer code.

Your view of the "best case XLS enhancement" (limit 100 words)

It would be great to define many next functions, preferably with some additional information that will help identify the test cases in the test summary. Some changes to the procs syntax were proposed in this comment, but the example given doesn’t show if specifying multiple test cases for the same design would be possible.

Here are some options for adding multiple test cases to DSLX:

1. Custom attribute for the next() function

#[test_proc]
proc PassthroughTest {
    terminator: chan<bool> out;
    data_s: chan<u32> out;
    data_r: chan<u32> in;

    config(terminator: chan<bool> out) {
        let (data_s, data_r) = chan<u32>("data");
        spawn Passthrough(data_r, data_s);
        (terminator, data_s, data_r)
    }

    #[test_case(name="First test case")]
    next() {
        ...
        send(tok, terminator, true);
    }

    #[test_case(name="Second test case")]
    next(count: u32) {
        ...
        send(tok, terminator, true);
    }
}

2. Custom attribute with test case name derived from a function name

#[test_proc]
proc PassthroughTest {
    terminator: chan<bool> out;
    data_s: chan<u32> out;
    data_r: chan<u32> in;

    config(terminator: chan<bool> out) {
        let (data_s, data_r) = chan<u32>("data");
        spawn Passthrough(data_r, data_s);
        (terminator, data_s, data_r)
    }

    #[test_case]
    case_one() {
        ...
        send(tok, terminator, true);
    }

    #[test_case]
    case_two(count: u32) {
        ...
        send(tok, terminator, true);
    }
}

3. Test cases naming convention taken, for example, from Google Test.

#[test_proc]
proc PassthroughTest {
    terminator: chan<bool> out;
    data_s: chan<u32> out;
    data_r: chan<u32> in;

    config(terminator: chan<bool> out) {
        let (data_s, data_r) = chan<u32>("data");
        spawn Passthrough(data_r, data_s);
        (terminator, data_s, data_r)
    }

    test_case_one() {
        ...
        send(tok, terminator, true);
    }

    test_case_two(count: u32) {
        ...
        send(tok, terminator, true);
    }
}

One of the challenges for enabling multiple test cases is how to write the init() function for each of them. However, I believe that in general state for test procs is not required at all, since the next() function can be replaced with a for loop.