Multiple definition of mapSecondGetter function

[daniele77]

An inline is missed in the global lambda function mapSecondGetter. Without it, a linker error is produced when the header file superior_mysqlpp.hpp is included in more than one translation unit

[daniele77]

On older compilers, the inline directive does not work on auto lambdas. This is the right code, instead:

template <typename T>
T& mapSecondGetter(T&& item) { return std::forward<T>(item).second; }

 // auto mapSecondGetter = [](auto&& item) -> auto& { return std::forward<decltype(item)>(item).second; };

[daniele77]

close #132

[OpatrilPeter]

Thank you for your interest and bringing it to our attention.

template T& mapSecondGetter(T&& item) { return std::forward(item).second; }

Firstly, that doesn't look right to me - item and result types are independent - in fact, T is pair while result type is second element. Following should probably work in C++14.

template <typename T>
auto& mapSecondGetter(T&& item) { return std::forward<T>(item).second; }

Secondly, lambda with auto in parameter ought to be templated implicitly, isn't the problem actually with the "global variable"? Using const auto instead of auto on the lambda should fix the problem, as global constants are implicitly static. I don't have GCC 10 at hand, so I can't test - if adding const won't help, posting the error message would be helpful.

[daniele77]

Thank you. Using const auto fixed the issue (with GCC 10). I updated the PR. Do you think the fix will be available in the next release?

[OpatrilPeter]

Thank you for your patience. I'll merge the changes and release a new version soon (note that we have recently discontinued OS-specific releases (such as deb packages), as they made little sense for header-only library).

As we've also changed our policy on commit messages, I went ahead and recreated the changes in different PR, so I'll be closing this one.