[math] TComplex arithmetic operators don't work with integral types

[meiyasan]

Feature description

Hello,

This sounds obvious but, can we hope to have some cast compatibility with int and float for TComplex class ?
Just typing the following line is failing: 2*TComplex(2,2)
but 2.0*TComplex(2,2) is fine.

I dont think this would be a lot of work, but very helpful to use.

Cheers,
M.

[couet]

I confirm. On Mac we get:

root [0] 2.*TComplex(2,2)
(TComplex) (4,4i)

root [1] 2*TComplex(2,2)
ROOT_prompt_1:1:2: error: use of overloaded operator '*' is ambiguous (with operand types 'int' and 'TComplex')
2*TComplex(2,2)
~^~~~~~~~~~~~~~
/Users/couet/git/couet-root-bin/include/TComplex.h:88:20: note: candidate function
   friend TComplex operator *(Double_t d, const TComplex & c)
                   ^
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, float)
2*TComplex(2,2)
 ^
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, double)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, long double)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, int)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, long)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, long long)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, __int128)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, unsigned int)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, unsigned long)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, unsigned long long)
ROOT_prompt_1:1:2: note: built-in candidate operator*(int, unsigned __int128)
root [2] 

[vepadulano]

For my own curiosity, what is the advantage of using TComplex over std::complex?

[guitargeek]

Probably compatibility with other ROOT interfaces. But indeed I don't think we need to support this TComplex class that much. I have opened a PR to address this issue though, so it can be closed.

[meiyasan]

Thank you I will follow your thread.

[guitargeek]

You're welcome! Let's keep this issue open as long as the PR is not merged though :)

[Axel-Naumann]

Alternatively and / or additionally it might make a lot of sense to remove the lines

   operator Float_t  () const {return static_cast<Float_t>(fRe);}
   operator Int_t    () const {return static_cast<Int_t>(fRe);}

and keep only the conversion to double.

[guitargeek]

I agree that this should be considered in the future. Implicit type conversions always come back to bite you at some point and should be forbidden 🙂