On 5 February 2013 10:20, Jonathan Wakely wrote: > On 5 February 2013 10:15, Jonathan Wakely wrote: >> On 5 February 2013 10:09, Jonathan Wakely wrote: >>>> Is this a bug in the 4.7.1 compiler or did the compiler change to conform >>>> to some standard that does not allow vector's to initialized to NULL. >>> >>> I don't think it changed in quite some time. It's a very well known >>> problem, see http://open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#438, >>> but the implementation is supposed to ensure it works. I can't >>> remember if we have an open bug about it still. >> >> Looking into it further, I think we conform to C++ 2003 + DRs and the >> workarounds given earlier are necessary. >> >> In C++ 2011 the implementation should accept the code, so we'll have >> to fix that. > > This is the DR we don't implement yet: > http://open-std.org/jtc1/sc22/wg21/docs/lwg-defects.html#1234 It's obviously too early in the morning for me, we've already fixed it on the trunk: http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43813 So it's an unfortunate defect in the C++ 2003 standard that the code doesn't compile, but G++ is conforming to reject it. The C++ 2011 standard requires it to compile, and G++ 4.8 will accept it in C++11 mode.
