[ruby-core:105668] [Ruby master Bug#18187] Float#clamp() returns ArgumentError (comparison of Float with 1 failed)

["mrkn (Kenta Murata)" <noreply@ruby-lang.org>]

Issue #18187 has been updated by mrkn (Kenta Murata).


I think it's OK to return NaN for all the cases of `Float::NAN.clamp`.

----------------------------------------
Bug #18187: Float#clamp() returns ArgumentError (comparison of Float with 1 failed)
https://bugs.ruby-lang.org/issues/18187#change-94167

* Author: SouravGoswami (Sourav Goswami)
* Status: Open
* Priority: Normal
* ruby -v: ruby 3.0.2p107 (2021-07-07 revision 0db68f0233) [x86_64-linux]
* Backport: 2.6: UNKNOWN, 2.7: UNKNOWN, 3.0: UNKNOWN
----------------------------------------
When I have a Float::NAN as a number, I expect all the method to work properly.

For example, `Float::NAN - 1` gives NAN. But Float::NAN.to_i raises FloatDomainError.

But in case of clamp(), Float::NAN.clamp(0, 100) returns `ArgumentError (comparison of Float with 1 failed)`

This error doesn't explain what's actually wrong. I didn't write the comparison to compare Float with 1. I didn't pass any invalid argument either. This error is a reflection of what's going on in the C level, which shouldn't appear to the user.

If I write a vanilla clamp() in ruby:

```
Float.define_method(:clamp2) { |min, max| self < min ? min : self > max ? max : self }
```

In this case, I can call it like this:

```
> 8.0.clamp2(10, 100)
=> 10
> 80.0.clamp2(10, 100)
=> 80.0
> 800.0.clamp2(10, 100)
=> 100
> Float::NAN.clamp2(10, 100)
=> NaN
```

As you can see, it just returns NAN. But in case of the built-in clamp, it raises the ArgumentError, even though my arguments are just correct. So this should handle this clamp() correctly, either returning the min value or `Float::NAN`.



-- 
https://bugs.ruby-lang.org/