If your premise is that size_t should never be used in an API, then I disagree. It is unfortunate that it causes mangling differences, but this is hardly the only package to have to deal with this issue.
The API is simply using size_t where the underlying STL containers use it too, which seems sensible.
If your premise is that size_t should never be used in an API, then I disagree. It is unfortunate that it causes mangling differences, but this is hardly the only package to have to deal with this issue.
The API is simply using size_t where the underlying STL containers use it too, which seems sensible.