On Tue, 17 Feb 2026 20:40:59 GMT, Andy Goryachev <[email protected]> wrote:
> OK, I am not against doing this in principle. But it's a new, non-trivial > algorithm that needs to have additional tests and the implementation needs to > be reviewed. The PR description should also change to reflect this added > complexity. The tests are there to proof that it parses doubles correctly. The algorithm can be treated as a black box. It doesn't need to be exhaustively checked, that has already been done far more thoroughly than we can hope to do -- it's like having to prove that some complex sorting algorithm works; you don't need to fully understand the algorithm, just verify some edge cases, to proof you didn't make a mistake "copying" the proven algorithm. I checked what Michael copied, and I didn't see any mistakes. The test cases (and probably numerous dependent tests in the rest of FX) proof that it works. ------------- PR Review Comment: https://git.openjdk.org/jfx/pull/2069#discussion_r2820987766
