We have a definition of the symbol gethostbyname_r in libc. It is problematic because:
1. It is not compatible with the glibc definition of the symbol. 2. It is not declared anywhere in a header file we install. So autoconf will detect that the symbol exists, a C compiler with -Wno-error will invent a bogus implicit declaration for the symbol, and there will be mysterious run-time misbheaviour because the calling code doesn't match the expectations of the gethostbyname_r definition. As far as I can tell, it has *never* been possible to legitimately use our gethostbyname_r symbol. The above scenario has already caused trouble in various pkgsrc packages. See, for example, <https://mail-index.netbsd.org/netbsd-users/2016/06/30/msg018651.html>. Can we remove gethostbyname_r now, before the mythical libc bump? Since it has never been possible to use the symbol legitimately, any application that *would* break if we removed it is *already* broken. The only difference is that if we remove the symbol from libc, then we get an immediate report that the application is broken, instead of mysterious run-time misbehaviour some time in the future. The same goes for the symbols: endhostent_r gethostbyaddr_r gethostbyname2_r gethostbyname_r gethostent_r sethostent_r
