Question

Can one generally assume thread-safety for get-methods if not otherwise noted in the documentation of a method? Or is it the other way round that one should never assume thread-safety if not otherwise noted? What do you think?

Edit:

Assume a class

class InfoClass {
  public:
    void Init();
    int  GetInfo();
    void Free();
};

Calling Init() once, GetInfo() from a threaded context, and Free() after the threaded operations have ended.

Answer 1

Thread-safety always requires extra work, so the only thing you can safely assume is the opposite – in the absence of a direct support the thread-safety is not supported.

Explanation:

Suppose you have a getter for a simple 64-bit long long and are running on a 32-bit architecture. While the computer is fetching the second half of that long 64-bit value (having just been done with the first half) another thread updates that second half and now what you have is an inconsistency – thus it’s not thread-safe.

Edit (to match the edit in the question):

(side note – the way you presented your class is making it unusable as all the members are private)

If you do not have any access methods that change the state of your class after it’s been built then you can assume thread-safety. But it’s still a sloppy slope as later on someone who doesn’t know about your assumption may add a setter to the class and have a wonderful journey in a random-error debugging experience 😉