Leif Hedstrom added a comment - 19/Feb/12 04:22 I got most of it to compile, with the exception of this in IpMap.cc: IpMap.cc:850:11: error: call to function ' operator <' that is neither visible in the template definition nor found by argument-dependent lookup if (x < n->_min) n = left(n); ^ IpMap.cc:1236:24: note: in instantiation of member function 'ts::detail::IpMapBase<ts::detail::Ip6Node>::contains' requested here zret = _m6 && _m6->contains(ink_inet_ip6_cast(target), ptr); ^ IpMap.cc:1164:13: note: ' operator <' should be declared prior to the call site or in the global namespace inline bool operator <(sockaddr_in6 const * lhs, sockaddr_in6 const & rhs) { ^ IpMap.cc:665:17: error: call to function ' operator ==' that is neither visible in the template definition nor found by argument-dependent lookup if (n->_min == min) { ^ IpMap.cc:1260:21: note: in instantiation of member function 'ts::detail::IpMapBase<ts::detail::Ip6Node>::mark' requested here this ->force6()->mark(ink_inet_ip6_cast(min), ink_inet_ip6_cast(max), data); ^ IpMap.cc:1172:13: note: ' operator ==' should be declared prior to the call site or in the global namespace inline bool operator ==(sockaddr_in6 const & lhs, sockaddr_in6 const * rhs) { ^ IpMap.cc:736:19: error: call to function ' operator <=' that is neither visible in the template definition nor found by argument-dependent lookup if (n->_max <= max_plus) { // completely covered, drop span, continue ^ IpMap.cc:1188:13: note: ' operator <=' should be declared prior to the call site or in the global namespace inline bool operator <=(sockaddr_in6 const & lhs, sockaddr_in6 const & rhs) { ^ IpMap.cc:514:16: error: call to function ' operator <' that is neither visible in the template definition nor found by argument-dependent lookup if (target < n->_min) n = left(n); ^ IpMap.cc:640:16: note: in instantiation of member function 'ts::detail::IpMapBase<ts::detail::Ip6Node>::lowerBound' requested here N* n = this ->lowerBound(min); // current node. ^ IpMap.cc:1260:21: note: in instantiation of member function 'ts::detail::IpMapBase<ts::detail::Ip6Node>::mark' requested here this ->force6()->mark(ink_inet_ip6_cast(min), ink_inet_ip6_cast(max), data); ^ IpMap.cc:1164:13: note: ' operator <' should be declared prior to the call site or in the global namespace inline bool operator <(sockaddr_in6 const * lhs, sockaddr_in6 const & rhs) { ^ 4 errors generated. I'll commit the other changes (after I test them on Linux), but I'll need assistant from some C++ propeller head with the template madness above .

Leif Hedstrom added a comment - 19/Feb/12 21:27 This fixes the build problems with clang. Alan, is this the right solution?

Igor Galić added a comment - 24/Feb/12 22:33 http://git-wip-us.apache.org/repos/asf?p=trafficserver.git;a=commit;h=23af5c5b56d8d3c4d61ace903908f98ec3b7d12a http://git-wip-us.apache.org/repos/asf?p=trafficserver.git;a=commit;h=fe1da80b94d368343174ada8b137a0c4ae1c387e http://git-wip-us.apache.org/repos/asf?p=trafficserver.git;a=commit;h=56c29c0a1dd950eba82039f3f4dcb950a1fdff6b

Igor Galić added a comment - 11/Apr/12 16:53 assigning to brian, since he's taking care of 3.0.x now

Jan-Frode Myklebust added a comment - 12/Apr/12 11:14 Brian, please see patch in TS-1185 . I've cherry picked commit 9e3459a1006ad2e4381d20fe5374a11994dccf88 and cea3c71360a066a24b39e36ee116e83ea1db1bc8 into v3.0.4 to make it build on fedora-17 which has gcc-4.7. It would be great if you could include these patches in the upcoming v3.0.5 release.

Brian Geffon added a comment - 12/Apr/12 20:39 - edited Ok, i'll make the final vote for it and it will be included in 3.0.5.

Brian Geffon added a comment - 13/Apr/12 00:02 @Jan-Frode Myklebust: all that will be included in 3.0.5 will be the commits referenced in the STATUS file which is only fe1da80b94d368343174ada8b137a0c4ae1c387e. @Igor, should there be multiple SHAs in the STATUS file, should I be backporting based on your comment above / and or start a new vote?

Igor Galić added a comment - 13/Apr/12 00:38 Brian, I only added the SHAs which seemed sensible. The one for configure.in to add -Werror didn't seem useful for this particular bug.

Brian Geffon added a comment - 26/Apr/12 03:05 Backported to 3.0.x in commit 944d5babae425706e6238904ee2df621a68ef010