System.NotImplementedException: byref delegate about ironpython3

[slozier]

If you don't mind I'll re-open the issue since this is technically a regression. I'm thinking a try/catch of NotImplementedException would do the trick (although not sure how I'd go about testing it)...

private TryGetProperty TryGetByJit(MethodInfo method) {
    try {
        return method.DeclaringType!.IsAssignableFrom(typeof(TSelfType))
            ? TryGetByJitWithoutCast(method)
            : TryGetByJitRequiringCast(method);
    } catch (NotImplementedException) { // Mono can throw this...
        return TryGetByReflection;
    }
}

[BCSharp]

Cannot reproduce, the Python sample is incomplete. What are arguments to set_generated_rules? Which version of .NET are you using?

[thislooksfun]

Cannot reproduce, the Python sample is incomplete.

I'm not sure what you mean. That's the exact script I've been running that has the issue for me.

What are arguments to set_generated_rules?

I'm not actually calling set_generated_rules at any point yet, that's the next step. I'm just calling compiled.Execute() so I can attach the script to the scope so I can then get the function out to call it later (if there's an easier way to do that let me know, but I couldn't find one).

Which version of .NET are you using?

I'm using .NET 4.7.2.

[BCSharp]

OK, it wasn't clear to me at which place you were getting the exception. As I understand it now, it is compiled.Execute(scope); and not when calling set_generated_rules. Still, I cannot reproduce your issue. I tried .NET Framework 4.8 and .NET 6.0 — the script, as it is, executes just fine.

On a side note, your script uses TRUE rather than True? This variable will be undefined unless you define it explicitly before calling set_generated_rules if you get that far (though it should not have any consequence on the problem you are facing now).

I'm using .NET 4.7.2.

Not using Mono? Strange that I see System.Reflection.MonoMethod.Invoke in the stack trace.

[thislooksfun]

OK, it wasn't clear to me at which place you were getting the exception. As I understand it now, it is compiled.Execute(scope); and not when calling set_generated_rules.

Correct, sorry I didn't make that clear at first.

Still, I cannot reproduce your issue. I tried .NET Framework 4.8 and .NET 6.0 — the script, as it is, executes just fine.

Huh. Weird. I'll do some more digging to try to find a better MRE.

On a side note, your script uses TRUE rather than True? This variable will be undefined unless you define it explicitly before calling set_generated_rules if you get that far (though it should not have any consequence on the problem you are facing now).

Good to know. I don't know python. That being said I'm sourcing the final script from elsewhere, so I'm not too concerned if my test script doesn't work 100%.

Not using Mono? Strange that I see System.Reflection.MonoMethod.Invoke in the stack trace.

I'll be honest, I have no idea. I'm not a C# dev, I'm just trying to make a mod for a Unity game. All I know is my .csproj says <TargetFramework>net472</TargetFramework>.

[BCSharp]

Ah, Unity… Just to make sure: are you running your mod on Unity on a desktop platform? If so, which one? Windows, macOS, UWP, Linux?

The "TargetFramework" moniker in .csproj specifies the .NET API, not necessarily the .NET platform. net472 means it is either .NET Framework or Mono. If you run on something else than Windows, then for sure it's Mono. Unfortunately, Mono does not implement the complete .NET API. Sometimes workarounds are possible.

[thislooksfun]

I'm developing the mod on Windows, but the game itself (Hollow Knight) is cross platform. Also, after doing some more digging through the project setup I see references to Mono.Cecil, so I'm pretty sure it's Mono.

That being said, I don't understand why Mono would be a problem here. The error being thrown looks like a missing/bugged IronPython language feature error to me (but one I can't find where it's thrown from), not a .NET vs Mono error. Though admittedly I know almost no C#, so it's possible I'm overlooking something.

[BCSharp]

It does look like a something in Mono not being fully implemented in the version of Mono you are using. I have found a similar report in mono/mono#21534. This issue is still open, however I was able to run your example without problems with Mono v6.12 on Linux, so maybe it got implemented anyway and you are simply using an old Mono version.

I haven't worked on any Unity mod before, so I don't know anything about Unity toolset, but if it is possible, I'd recommend to migrate your mod project to .NET 6.0 or later. If you point me to some resources on Unity mod development in C#, I might be able to look into it (I have Unity on Windows installed).

[thislooksfun]

Ah, I see. Well that's a shame. Unfortunately I don't think I have a way to adjust the Mono version used. I'm merely creating a dll that is then loaded by a Unity game outside of my control. I'll keep poking around though, see if there's some other solution I can find.

[slozier]

@thislooksfun Does your code fail with v3.4.0? You're hitting a code path that was recently modified for performance (#1706) so if v3.4.0 works it might be possible to change some stuff to make it work again...

[thislooksfun]

It does! Thank you!

However I now have a new problem in that now calling scope.GetVariable() is throwing System.TypeLoadException: Failure has occurred while loading a type., but that feels like a problem with my code rather than the library.

[BCSharp]

not sure how I'd go about testing it

.NET is being tested by several test runs on all supported versions, maybe something similar for Mono: one test run on the latest Mono version and one on the oldest that we support (though I don't know which one that is).

[slozier]

Well in theory the ".NET Framework" tests on Linux/macOS are Mono. Not sure which version, but I'm assuming they're on 6.12.something as that's the latest? As for the oldest that we should support probably depends on whats bundled with Unity as that appears to popular among users.