Thanks for the clear and minimal repro @patrykpiotrmarek
I'm not immediately sure what the cause may be here so it could take a bit of investigation from the team. Would you be able to upload your repro to https://github.com/nunit/nunit.issues ?

A repro for this was uploaded by @patrykpiotrmarek for investigation in nunit/nunit.issues#3

Interesting, and fascinating. I can confirm this from within Visual Studio using the "Tests" output pane.
On 3.14.0 I can see the output yet on 4.0.0, 4.0.1, and 4.1.0 I can't. Something in 3.14.0 -> 4.0.0 changes this behaviour.

Hm. Ok. This seems to manifest on net8 only. I can still see the output on net48 when consuming NUnit 4.0.0 or 4.1.0

I can't really explain why it happens, but this feels related to the changes in the Timeout attribute.

This setting gets read in here:

Which I think then causes a TimeoutCommand to be created instead of a CancelAfterCommand here:

Adding CancelAfter on the affected class does indeed cause the output to reappear as the presence of the property will instead cause the CancelAfterCommand to be used instead there.

// This works
[TestFixture, CancelAfter(60000)]
public class Tests
{
    [Test]
    public void Test1()
    {
        TestContext.WriteLine("line1");
        Assert.That(1, Is.EqualTo(0));
    }

    [TearDown]
    public void TearDown()
    {
    }
}

@patrykpiotrmarek Sorry it took us a while to investigate. This means that at least as a workaround it should be possible to see the output on .NET 8 by applying the CancelAfter attribute on the classes of any affected tests.

Note that CancelAfter is using co-operative multi-tasking. It will not abort your tests unless you obey the CancellationToken.

Good point, thanks @manfred-brands ! I had forgotten to mention that.

It's also worth mentioning that CancelAfter is a better alternative to Timeout on newer runtimes like NET8 in particular if a test does out of process or generally "async" work as the cooperative cancellation model facilitates the correct cancellation of async calls.

A bit of an update here as I have to pause my own investigation for now

An early option I'd considered but had to rule out was passing the TextWriter around by simply adding a new parameter to MakeTestResult(). As a public abstract method, it was a challenge to do this in a way which didn't have an outside chance of breaking a consumer somewhere. I opted for caution here as this appears a common integration point.

I did manage to "fix" it in a branch here: https://github.com/stevenaw/nunit/tree/4598-timeout-context-output though I'm not particularly happy with the result. It meant extending and wrapping a TextWriter as a publicly exposed property and changing a readonly field to be internally settable. This felt a little heavy-handed given this may become a .NET6-only issue if we did the Timeout rework in #4673.

An idea I haven't time to pursue yet is "snapshotting" the state of the TestResult at the time of the Timeout and applying it to the context here. It's possible there may be other options here too.

@stevenaw the timeout refactor is working good for net462 and net8. For net6 this now throws a PlatformNotSupported exception as it never previously worked as intended. I hope to get the PR up this weekend.

Fantastic, thanks @manfred-brands .
I took a quick look through your Timeout PR and it looks like it's taking things in a great direction