On 2020/05/10 09:29:48, hanwenn wrote: > https://codereview.appspot.com/566080043/diff/560020046/lily/freetype.cc > File lily/freetype.cc (right): > > https://codereview.appspot.com/566080043/diff/560020046/lily/freetype.cc#newcode143 > lily/freetype.cc:143: }; > On 2020/05/10 09:16:58, hahnjo wrote: > > Not sure if FT developers plan to change this interface at some point. In > other > > projects, I have seen something akin to > > FT_Outline_Funcs funcs; > > memset(&funcs, 0, sizeof(funcs)); > > for external structs from dependencies to make sure no field goes > uninitialized. > > the C99 standard actually says that omitted fields are zero-initialized > > https://gcc.gnu.org/onlinedocs/gcc/Designated-Inits.html > > I added the field because GCC warns about it (I don't understand why)
It could be that this is because of C++ officially inheriting from C89 IIRC? https://en.cppreference.com/w/cpp/language/aggregate_initialization says: "If the number of initializer clauses is less than the number of members [...] the remaining members [...] are [...] copy-initialized from empty lists, in accordance with the usual list-initialization rules (which performs value-initialization for non-class types [...])" which I would read as "uninitialized". > https://codereview.appspot.com/566080043/diff/560020046/lily/freetype.cc#newcode149 > lily/freetype.cc:149: return ((Path_interpreter *) user)->moveto (*to); > On 2020/05/10 09:16:59, hahnjo wrote: > > Do you want to create copies of the arguments? (here and for the other > > functions) > > yes. I find it easier to reason about and read. It's all inlined anyway. const refs should work the same AFAICS https://codereview.appspot.com/566080043/
