Utilise iterators instead of std::size_t about fuzzylite HOT 3 CLOSED

Hi Juan,
I would also suggest to use the std::size_t type in place of the int type in methods like

• int numberOfXXX(),
• ... getXXX(int),
• void setXXX(..., int)
• ... removeXXX(int)

Some weeks ago, I prepared a specific branch in my fork

https://github.com/sguazt/fuzzylite/tree/patch/use-size_t

which however was based on a fork of your v5.x branch.
Currently I have no time for development, but maybe in the next months I can try to port the changes in the develop branch.

Let me know.
Cheers,

Marco

from fuzzylite.

Hi Marco,

thank you very much for your work! I will check if I can merge that into the develop branch.

However, could you elaborate a bit on why should I use return types of std::size_t instead of std::vector<T>::size_type? I really liked int because it is less verbose, but if this needs to be changed, then why not the std::vector<T>::size_type?

Cheers,
Juan.

from fuzzylite.

Hi,
The reason to use std::size_t in place of int is to avoid possible conversion errors.
Most of the containers used inside fuzzylite classes use an unsigned integral type to return the number of elements, while the int numberOfXXX() methods use a signed integral type. This means that if the number of elements stored in the container is greater than std::numeric_limits<int>::max() it is likely to run into an overflow problem.
Also note that std::size_t is usually implemented as unsigned long, so you got an additional problem in case sizeof(int) < sizeof(long).

Regarding your proposal to use std::vector<T>::size_type, personally I don't like it since:

1. You expose to the user of the classes the internal details of your implementation. If one day, for some reason, you decide to use a container other than std::vector you'll have to change the public signature of your APIs to match the new_container::size_type type.
2. As far as I know. the type std::size_t is really the standard and recommended way (in the C++ sense) to refer to types that represents a size (likewise, std::ptrdiff_t is the standard type to use to represent differences and offsets). From the C++14 draft standard (sec. 18.2):
   • The type size_t is an implementation-defined unsigned integer type that is large enough to contain the size in bytes of any object
   • The type ptrdiff_t is an implementation-defined signed integer type that can hold the difference of two subscripts in an array object

Obviously, this is my personal point of view

Cheers,

Marco

from fuzzylite.