On 19.04.2010, at 10:19, Duncan Gibson wrote: > Ian: >> I was going to say that the wchar_t thing was likely to give us >> problems on win32, but I see Albrecht already got there. > > Yes, I didn't know whether to go for the typedef, or the #define. > As the scope of the latter is usually cited as one of the big > deficiencies of C/C++, I thought the typedef would be better. > It would appear not.
In general, typedef would be better, IMHO, but here it appears that we don't have a choice, because we would have to override another typedef that has been defined before. [ BTW.: Is there a way to "undefine" a typedef ? ] In fact, we're not changing the typedef, but we're "changing" MK's code to use unsigned instead of wchar_t, thus making the typedef irrelevant. That said, this is IMHO a hack that doesn't make the code transparent for readers: who would know that mk_wcwidth.c is only included in another c file, and that *this* file defines wchar_t to be something completely different, if he sees only mk_wcwidth.c ? Thus, I agree with Ian: since we change it anyway, we should probably do it the clean way and change the code directly, or add the #define in mk_wcwidth.c at least. Albrecht _______________________________________________ fltk-dev mailing list [email protected] http://lists.easysw.com/mailman/listinfo/fltk-dev
