Carl Dehlin

I was not aware that a decision had already been made to change minimum required version to 3.5 (PR here: https://github.com/doctest/doctest/pull/811). Then indeed this is a non-discussion, just a new...

Thanks for your input @a4z . Then it does sound like the cmake strategy is to continuously bump up the minimum required version in order to stay forward compatible with...

@Saalvage Should ping you too ofc! See the TLDR above

@tcbrindle This looks promising. Maybe I misunderstood how the `[...max]` syntax works in CMake, but doesn't that document the maximum version of CMake your code builds? As I understand it...

I agree with everything you have described, and believe the current behavior is not intentional. Since the current output is completely broken when using test suites I think it's fine...

JUnit5 also supports "legacy XML": https://junit.org/junit5/docs/current/user-guide/#junit-platform-reporting-legacy-xml Open Test Reporting support two different schemas as well: - Event based: https://github.com/ota4j-team/open-test-reporting/?tab=readme-ov-file#event-based-format - Hierarchical: https://github.com/ota4j-team/open-test-reporting/?tab=readme-ov-file#hierarchical-format I agree, need to pick a schema. Question...

What is your doctest version? This works just fine in >= 2.4.11 at least

Oh now I see, I just took your code, fixed the errors and compiled it. I'm curious, why not just ```cpp constexpr auto my_func(auto val) { REQUIRE(0LL < val); }...

It seems that the `ImageFeatures` is an internal implementation detail? It's in the `cv::detail` namespace, I wouldn't expect it to appear in the pybindings to begin with.

Do you mean that the errors for 0 and 1 dimensions are expected? I feels like a very bad bug to me, and it will leave ONNX integration in a...