Modernize NULL with nullptr in newview #5107
Conversation
| LPCWSTR unique_mutex_name = L"SecondLifeAppMutex"; | ||
| HANDLE hMutex; | ||
| hMutex = CreateMutex(NULL, TRUE, unique_mutex_name); | ||
| hMutex = CreateMutex(nullptr, TRUE, unique_mutex_name); |
There was a problem hiding this comment.
Bulk searched & replaced? Windows API calls should IMHO keep using NULL.
There was a problem hiding this comment.
I'm curious about the reasoning for this. Not saying you are wrong but would like to learn the possible pitfalls with this.
There was a problem hiding this comment.
NULL equals to 0 and I don't think we should mess with the expected values of the Windows API. This and the other PR just look like a bulk search and replace change without taking OS-specific API implementations into account.
It is kinda weird this and the other PR use 3P compatibility as a reasoning for this change, but then completely disregards OS-specific API implementations. This line is just a placeholder for all changes using OS-specific API call and there a for sure a bunch more of these cases. In this particular case, you can even notice that TRUE hasn't changed to true for exactky that reason.
I don't dislike or dismiss the idea of this PR in general, but it seems like a bulk search and replace PR without taking individual cases into account.
There was a problem hiding this comment.
Can't argue with this. It probably is a bulk replace. What I wondered is, does nullptr not resolve to a the same actual 0x0000000000000000 value being passed on the stack? Maybe in certain compiler backends or whatever?
I can understand that the NULL was always a bit of a hack that was there because of the lack of a compiler native definition, but I would expect both to resolve ultimately to the same code unless there is some library specific optimization that a compiler "might" employ. For the WinAPI I would expect a compiler to stay far from trying to do any such silly things if it does not want to break anything in spectacular ways. But I'm learning here and just as with the STL being still quite an unfamiliar thing until now, I also am far from being even a C++ specialist.
Maybe there exist compilers that would bark on nullptr being passed to a standard C pointer of anything? Would seem weird but I'm not claiming it could not exist.
3 participants
Description
Modernize NULL with nullptr in newview for increased type safety and compatibility with modern c++ library expectations that compare to nullptr_t but have no compatibility for NULL.
Related Issues
Issue Link: #5078
Checklist
Please ensure the following before requesting review:
Additional Notes