Removing the code with less precision and uncommenting the code with …

[lgahdl]

…more precision

[johngrantuk]

This does seem promising! It also fixes the consistently failing example I have. You'll need to investigate the regression with the failing tests though. Let us know how you get on!

[lgahdl]

This does seem promising! It also fixes the consistently failing example I have. You'll need to investigate the regression with the failing tests though. Let us know how you get on!

Since all tests failed with hardcoded values, and the result values were very close to the hardcoded one, I changed the hardcoded values, let me know if I should take another approach

[brunoguerios]

Nice that you found the issue 👏
Hardcoded values do seem pretty close - I'd say updating them so tests pass should be ok 👍