Treat rc == 0 as success?

[AntonDevil]

Hi @StackOverflowExcept1on

You're the boos here.

net-core-injector/Bootstrapper/src/library.cpp

Lines 81 to 86 in 9b456fd

	/// Success_HostAlreadyInitialized = 0x00000001
	/// @see https://github.com/dotnet/runtime/blob/main/docs/design/features/host-error-codes.md
	if (rc != 1 || ctx == nullptr) {
	hostfxr_close_fptr(ctx);
	return InitializeResult::InitializeRuntimeConfigError;
	}

Can you explain why we treat rc == 0 as failure?
I don't get why it won't be ok to treat rc == 0 as success.
Maybe leave a comment about it?

[StackOverflowExcept1on]

I think because if you inject into application, you expect the context to already exist, right? I think this was related to dotnet/runtime#77091.

[AntonDevil]

Yes. I agree. Normally, I would expect the context to exist.

But what if returned rc == 0? This would mean that that the .NET context didn't exist before and it's our call initialized it, right? And we are "injected" our code at this point. Maybe the process we injected to was a kind of a launcher, that kicks off/bootstraps .NET app (but then that .NET app probably will be spawned as separate process...)

I'm not sure... I know that rc == 1 is what we would normally expect. But are there legitimate use cases where rc == 0 would also be valid case for successful injection? I can't think of one, but maybe you can? Or rc == 0 case is a tricky one and would case more problems than solve any if we allow it to succeed?