On 10/17/11 6:43 PM, "ext Oswald Buddenhagen" <[email protected]> wrote:
>On Mon, Oct 17, 2011 at 03:04:41PM +0200, Knoll Lars (Nokia-MP-Qt/Oslo) >wrote: >> On 10/17/11 1:35 PM, "ext Oswald Buddenhagen" >><[email protected]> wrote: > >> > i would just make Q_DECLARE_TYPEINFO for Q_LARGE_TYPE completely >> > explicit. >> >> Being able to change it is ok, but I do thing having a reasonable >> default is valuable. >> >the value for "reasonable" varies wildly between usage patterns: >appending/prepending often benefits from in-array allocation (saved >allocs/frees), while insertion/removal in the middle of a big list >benefits from out-of-array allocation (smaller memmoves; but then, if >this becomes a problem, you probably have the wrong data structure >anyway). > >> Otherwise everything's large by default, which would IMO lead to >> rather bad characteristics for simple&small user defined classes. >> >default to small. error out if the type is not primitive/movable. Bad idea. Most developers don't even know about our trait system, and neither should they. I can already imagine all the questions and bug reports we would get because of this. Lars _______________________________________________ Qt5-feedback mailing list [email protected] http://lists.qt.nokia.com/mailman/listinfo/qt5-feedback
