Tom St Denis <tstdenis@xxxxxxxxxxxxxxxx> writes: > Sergei Organov wrote: >> Tom St Denis <tstdenis@xxxxxxxxxxxxxxxx> writes: >> >>> Have you considered not passing a float? Or performing a cast? >>> >> >> Yes I had. I do want implicit conversion in some of those cases. >> >> I'd be happy to add a cast, but there is no way in C/C++ to say: "cast it >> to the actual type of the function argument". >> > Unless you're using third party code that continually changes (which > would be bad anyways) you should be able to look at the prototype and > work from that. It shouldn't be a mystery as to what a function > requires. Any alive code changes, being it first party, second party, or whatever. Implicit conversion just fits better in some cases. >> I've now checked GCC sources, and it seems that this warning is >> unconditional in 4.1.1, whereas it has been fixed and put under >> -Wconversion in 4.2.0. >> >> Yeah, I mostly use C++. >> > This is a function of the C++ language, No, it is not. > I wouldn't look to turning off the warning as a good thing as C++ is > supposed to be more type safe than C. A warning is just a warning. It gives false positives. If it were a mistake according to the language definition, the compiler would issue error. It's not an error in C++ to implicitly convert from float to int. -- Sergei.
