So if I compile MultiCache.cc with g++, then it works. So I can build traffic server on OS X by using g++ for MultiCache.cc and clang++ for everything else. :-)
I would imagine that it's easier to fix MultiCache.cc to not fail in clang++ then to work around the g++ compiler bug that is triggered by Diags.cc, but I'll leave that to the pros... Marc On Mon, Mar 26, 2012 at 8:12 AM, Marc Abramowitz <[email protected]> wrote: > trafficserver-3.1.3-unstable$ ./configure CC=clang CXX=clang++ && make > ... > clang++ -DHAVE_CONFIG_H -I. -I../../lib/ts -I../../iocore/eventsystem > -I../../iocore/net -I../../iocore/aio -I../../iocore/hostdb > -I../../iocore/cache -I../../iocore/cluster -I../../iocore/utils > -I../../iocore/dns -I../../lib -I../../lib/records -I../../lib/ts > -I../../proxy -I../../proxy/hdrs -I../../proxy/http -I../../mgmt > -I../../mgmt/utils -I../../mgmt/preparse -I/opt/local/include > -D_LARGEFILE64_SOURCE=1 -D_COMPILE64BIT_SOURCE=1 -D_GNU_SOURCE -D_REENTRANT > -Ddarwin -I/usr/local/include > -I/System/Library/Frameworks/Tcl.framework/Versions/8.5/Headers > -I/usr/local/Cellar/pcre/8.30/include -g -pipe -Wall -Werror > -Wno-deprecated-declarations -O3 -fno-strict-aliasing -Wno-invalid-offsetof > -MT MultiCache.o -MD -MP -MF .deps/MultiCache.Tpo -c -o MultiCache.o > MultiCache.cc > In file included from MultiCache.cc:34: > In file included from ./P_HostDB.h:52: > ./P_MultiCache.h:119:8: error: cannot initialize a parameter of type 'void > *' with an rvalue of type 'volatile int (*)[2]' > struct MultiCacheHeader > ^~~~~~~~~~~~~~~~ > MultiCache.cc:549:18: note: implicit default copy assignment operator for > 'MultiCacheHeader' first required here > *mapped_header = *(MultiCacheHeader *) this; > ^ > 1 error generated. > make[2]: *** [MultiCache.o] Error 1 > make[1]: *** [all-recursive] Error 1 > make: *** [all-recursive] Error 1 > > On Mon, Mar 26, 2012 at 7:15 AM, Leif Hedstrom <[email protected]> wrote: > >> On 3/25/12 10:33 PM, Marc Abramowitz wrote: >> >>> Trying to build 3.1.3 on OS X 10.6.8 and it seems to be triggering a >>> compiler error. Can anyone reproduce this? >>> >>> libtool: compile: /usr/bin/llvm-g++ -DHAVE_CONFIG_H -I. -I../../lib >>> -I/opt/local/include -D_LARGEFILE64_SOURCE=1 -D_COMPILE64BIT_SOURCE=1 >>> -D_GNU_SOURCE -D_REENTRANT -Ddarwin -I/usr/local/include >>> -I/System/Library/Frameworks/**Tcl.framework/Versions/8.5/**Headers >>> -I/usr/local/Cellar/pcre/8.30/**include -Os -pipe -march=core2 -msse4 >>> -Wno-deprecated-declarations -g -Wall -Werror -feliminate-unused-debug-* >>> *symbols -fno-strict-aliasing -Wno-invalid-offsetof -MT Bitops.lo -MD >>> -MP -MF .deps/Bitops.Tpo -c Bitops.cc -fno-common -DPIC -o .libs/Bitops.o >>> Diags.cc:547: internal compiler error: Segmentation fault >>> Please submit a full bug report, >>> with preprocessed source if appropriate. >>> >>> >> Yeah, try switching to the "real" llvm (clang / clang++) and not the old >> gcc wrapper. >> >> -- leif >> >> >
