[fsloth]

Ah, sorry.

So the reason _why_ we want the code to be explicit is that we want to be explicit always of the coordinate system we are in. Otherwise _extremely_ hard to find bugs can and will creep in - a float is a float.

The stuff I worked on input coordinates were left handed but the output coordinates were right handed.

[tantalor]

> a float is a float

That's not what is happening here. With this software there are no raw floats; all variables are typed by the kind of coordinate system (WGS84, ECEF, etc.)

Using your example, the variables would be typed by left- or right-handed. So it is impossible to mix them up, e.g., perform some illegal operation combining them.

Whether implicit conversion is allowed is just saying, does the programmer have to call a function to convert type A -> B, or can the compiler do that for them?

For example,

  void doFoo(LeftHandedVec3 v);

  RightHandedVec3 myData;

  // With explicit conversion.
  doFoo(convertToLeftHanded(myData));
  
  // With implicit conversion.
  doFoo(myData);
 

Does one of these produce more bugs?

[fsloth]

In my experience for technical applications

doFoo(convertToLeftHanded(myData));

wins in every way.

This is a very good example of ”code that is so obvious it looks stupid” - and which would be the main style I would promote.