Corentin Schreiber

> It is a desirable property because it is an _expected_ property of the comparison when dealing with numerical comparisons in literature. This is not the same as being intuitive...

So I think it all boils down to terminology; perhaps `Approx` was surprising because of its name, which reminds of the mathematical expression "x is approximately equal to y" (which...

Hi! There is currently no way to do this. Filtering a test case to only run certain sections would be possible, without too much cost, but figuring out the filtering...

For reference, *Catch2* does this with a [separate CLI option `--section ...`](https://github.com/catchorg/Catch2/blob/devel/docs/command-line.md#specify-the-section-to-run). If this is implemented in *snitch*, we will want to follow the same CLI API to benefit from...

> So if floating point std::to_char isn't available would you prefer to fall back to your constexpr function, snprintf, or something else? I think falling back to the constexpr functions...

The CI failures weren't related to this PR and have been fixed in `main`. Otherwise, I have benchmarked using `append_constexpr()` to implement `append_fast()`, the changes are in the [append](https://github.com/snitch-org/snitch/tree/append) branch....

> I think in the end it is your decision but since there is minimal performance difference I would lean towards using the constexpr version so that those functions get...

> Not sure why this wouldn't happen in debug, or normal clang. The offending number that fails the regex (1e-6, it seems) is the duration of a test (in seconds),...

Looking good! I'm happy to approve it once the small changes requested above are made.

This issue has reappeared in the current master branch. I noticed while trying to compile *doctest* to standalone wasm using Emscripten (standalone wasm doesn't have signals).