Ruben Bartelink

Right, what I was nudging at is that I suspect that it will work fine if you actually pass (an empty array of) arguments in the InlineData like one normally...

I agree that it is inconsistent behavior, and its good that you have provided a repro. As for whether its possible or important to fix, that's what I'm more pushing...

>Testing against an empty array of data is very basic when dealing with collection of data. I don't agree it is a pretty moot case. Hm, I missed the point...

Rather than disabling the test case, the workaround I'd suggest is to use `MemberData` to `yield return` each of the InlineData values you have; this should allow you to pass...

I agree in general, but [it can work given the right conditions](https://github.com/jet/FsCodec/blob/8f765f780eedb6aea88151fbd901e49bbac66e05/tests/FsCodec.NewtonsoftJson.Tests/VerbatimUtf8ConverterTests.fs#L113-L119)

You'll get quicker and better answers if you post this on stackoverflow.com; there are many more eyes on it there. I personally would read the code and/or the xmldocs for...

Is that the overload you're calling (maybe there is an `IEnumerable`-specific one that's winning, in which case adding a `` might make it work?) My point about the documentation was...

TL;DR the workaround is to reference a safe version for any indirectly referenced lib. Given you don't actually ship the tests and/or have them in a prod environment, it's arguably...

@artkamote Yes, it's less indirect, but 1. you can work around by using a later version 2. this is only for test code that should not be in prod

https://andrewlock.net/tracking-down-a-hanging-xunit-test-in-ci-building-a-custom-test-framework/ See also https://twitter.com/andrewlocknet/status/1508845968833363973 I really don't get the point in you logging this without a repro Unfortunately, the chances are SELECT is not broken, and you have debugging to...