On 1 March 2013 16:56, Daniel Hartwig <mand...@gmail.com> wrote: > On 1 March 2013 23:55, Dmitrijs Ledkovs <x...@debian.org> wrote: >> On 1 March 2013 14:17, Daniel Hartwig <mand...@gmail.com> wrote: >>> On 1 March 2013 21:14, Dmitrijs Ledkovs <x...@debian.org> wrote: >>>> In Ubuntu, instead of applying the full patch against 1.49, we went >>>> for a minimalistic hack/patch to workaround TIME_UTC in eglibc-2.16. >>>> Simply undef TIME_UTC if defined. >>> >>> Is that expected to work for a program that uses both libboost-thread >>> and C11 date and time? >> >> yes, as both have always evaluated to the same constant value on linux. > > On glibc, rather. An unsafe assumption, given that C11 does not > define the value any more than a non-zero, positive integer. Boost > does not even specify that much. > > These are different APIs, it is not precise to overload a single > macro. The boost macro exists in a distinct namespace and the Ubuntu > patch does not appear to have the feature you claim. From the > attached test case: > > $ g++ -o test -lboost_thread test.cc > test.cc: In function ‘int main(int, const char**)’: > test.cc:8:11: error: ‘TIME_UTC’ was not declared in this scope > int x = TIME_UTC; > ^ > -- > > Further, #undef is placed outside of the BOOST_XTIME_WEK070601_HPP > header guard; a second include of this header would undefine and not > subsequently redefine the macro. >
Thank you for explaining your point. You are right. >>> Is that expected to work for a program that uses both libboost-thread >>> and C11 date and time? existing code that uses libboost-thread does work with both old/new glibc. At the time that was what I tested and was important to keep. Indeed one doesn't get access to C11 date and time while using libbost-thread as patched in boost1.49 in ubuntu. Meh. I see little point of fixing this in ubuntu as C11 in gcc-4.7 is still not stable and is not encouraged to be used. Regards, Dmitrijs. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
