Created attachment 33869 [details] compile fix for ambigous call to abs() DateMath.cpp calls abs(long int) - which is ambiguous in some tool chains, for example on QNX: /home/harald/qt/qt/src/3rdparty/webkit/JavaScriptCore/wtf/DateMath.cpp: In function 'double WTF::parseDateFromNullTerminatedCharacters(const char*)': /home/harald/qt/qt/src/3rdparty/webkit/JavaScriptCore/wtf/DateMath.cpp:844: error: call of overloaded 'abs(long int&)' is ambiguous /opt/qnx640/target/qnx6///usr/include/stdlib.h:122: note: candidates are: int std::abs(int) /opt/qnx640/target/qnx6///usr/include/math.h:1042: note: long double std::abs(long double) /opt/qnx640/target/qnx6///usr/include/math.h:692: note: float std::abs(float) /opt/qnx640/target/qnx6///usr/include/math.h:624: note: double std::abs(double) make[1]: *** [obj/debug/DateMath.o] Error 1 Patch replaces the abs() call to labs(), which takes a long int and returns a long int. This also removes two implicit casts.
Comment on attachment 33869 [details] compile fix for ambigous call to abs() Another approach we use for issues like this is to put code in MathExtras.h to work around platform differences. Take a look at MathExtras.h to see what I mean. But this change seems OK. I think the change log is not so good. It says "fix compilation" which is far too vague. The bug mentions the QNX platform and the change log should too. I'm going to say r=me, though, because this seems like a good fix.
Comment on attachment 33869 [details] compile fix for ambigous call to abs() Please fix the ChangeLog per Darin's comments.
Created attachment 33970 [details] same patch, but with updated ChangeLog entry Updated ChangeLog entry as requested.
Comment on attachment 33970 [details] same patch, but with updated ChangeLog entry Clearing review flag on attachment: 33970 Sending JavaScriptCore/ChangeLog Sending JavaScriptCore/wtf/DateMath.cpp Transmitting file data .. Committed revision 46746. http://trac.webkit.org/changeset/46746
All reviewed patches have been landed. Closing bug.