On Friday 26 October 2012 17:04:33 Michael Pyne wrote: > On Friday, October 26, 2012 10:16:50 Frank Reininghaus wrote: > > > I was thinking the first one, but indeed Qt often does the second one, > > > I'm not sure what the difference really is. Either one is fine with me. > > > > in the end, I did not inline the function - before I pushed the > > commit, I wondered whether making a non-inline function inline is > > guaranteed to be binary compatible. According to Techbase, it's not: > > > > http://techbase.kde.org/Policies/Binary_Compatibility_Issues_With_C++#The_ > > Do .27s_and_Don.27ts > > More specifically, the non-inlined version needs to exist.
Oh OK, that sounds complicated for not much gain, you're right to have skipped this. > The compiler is > still within its rights to actually inline the code if possible (unless you > mark it as no_inline, which takes extra effort on your part). > > Or in other words, the compiler might do the right thing for you regardless. Might --- or might not :-) -- David Faure, [email protected], http://www.davidfaure.fr Working on KDE, in particular KDE Frameworks 5
