Buddy doesn't work on the C# target.

[mikkelmr]

Running this test on the C# target:

class TestParser extends buddy.SingleSuite {
    public function new() {
        describe("Test something", {});
    }
}

fails with the following error:

haxelib run hxcs hxcs_build.txt --haxe-version 4005 --feature-level 1
src\buddy\Should.cs(529,62): error CS0030: Cannot convert type 'buddy.ShouldInt' to 'buddy.Should<int>'
src\buddy\Should.cs(872,65): error CS0030: Cannot convert type 'buddy.ShouldFloat' to 'buddy.Should<double>'
Compilation error
Native compilation failed

[ciscoheat]

Hi, the C# target hasn't been well tested for the last couple of year(s), so I'm not surprised some issues has been introduced. I don't have too much time at the moment, so if you can give it a try and make a PR, that'll be much appreciated. If not I'll take a look when I get an opportunity.

[mikkelmr]

Fair enough! I word though appreciate it you could point me in a direction of where to look in the source. Just roughly, should I look north or south?

[ciscoheat]

Of course. :) The error seems related to should, and the should class is pretty much self-contained, so take a look and see if there are some casting issues there specifically for C#, maybe with the generics.

[mikkelmr]

Ok, took some time to dig into the problem and it seems to be related to ShouldInt(& ShouldFloat) class that have this construction:

@:keep class ShouldInt extends Should<Null<Int>> which causes the Haxe compiller to produce invalid C# code:
(That is Null<Int> as Type parameter does not work when extending a class with that as type parameter. Have reproduced in an isolated example, but since the C# target has very low priority I don't expect it to be fixed.)

If I change it to @:keep class ShouldInt extends Should<Int> then it compiles but then it's not possible to test for Nullable Int/Float types, so that's not a solution.
But maybe it would work if ShouldInt & ShouldFloat did not extended Should, but instead had the full implementation each. Not the most beautiful solution though.

Btw. it seems that in some cases the ShouldFloat class is used when testing types of Int - haven't quite figured out why that happens. Is this a known issue?