feat: add float32 and float64 support

[ccoVeille]

Fixes #7

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (f7927b7) to head (328cfff).
Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main       #14   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            1         1           
  Lines           54        60    +6     
=========================================
+ Hits            54        60    +6     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ldemailly]

it's a bit odd to merge something that only covers 1 case (float to uint) and not all the others? hopefully you don't tag this

[ccoVeille]

it's a bit odd to merge something that only covers 1 case (float to uint) and not all the others? hopefully you don't tag this

This PR allowed float32/64 to all integer types.

Were you talking about having a way to convert any int to float32/64. There is no overflow for these ?

As far as I know the only other overflow I'm not supporting is float64 to float32.

I might add them later.

Everything can be stored into a float64, so no overflow.

[ldemailly]

where is the check for overflow for say uint32

and none of the signed overflow are checked

[ccoVeille]

where is the check for overflow for say uint32

Are you talking about a bug in the code or a missing test?

Are you talking about the test I made when converting a float32/flpat64 to uint32?

Could you spot the exact line in the PR where I should have added something?

Maybe you are talking about the fact the code will cast math.MaxFloat32 with uint64 and compare it to math.MaxUint32

This might be the issue you were talking because yes, the code and tests might have a bug.

But I think refactoring in might address it #18, I may have to split the PR indeed

none of the signed overflow are checked

I noticed and I addressed it (I think) in #18

[ccoVeille]

Maybe you are talking about the fact the code will cast math.MaxFloat32 with uint64 and compare it to math.MaxUint32\n\nThis might be the issue you were talking because yes, the code and tests might have a bug.

Comfirmed https://go.dev/play/p/3DlJcb10eYs

But it's fixed via #18, I might reorganize the commits of this PR to fix the bug before refactoring to add the new error message

[ccoVeille]

Follow up will be in #23