Some additional comments: 1) You need to use -fopenmp=libomp to enable OpenMP in clang. -fopenmp just links libgomp but ignores all the pragmas. Weird, I know — and will be changed in the trunk soon. 2) 3.7 is the first version that supports OpenMP. 3.6 doesn’t. 3) clang is only able to work with libomp. … Read more
As others have mentioned, Annex D of ISO/IEC 9899:2011 lists the hexadecimal values of characters valid for universal character names in C11. (I won’t bother repeating it here.) I have been searching for an answer as to “why” this list was chosen. Character set standards First, there are two relevant standards defining a set of … Read more
How much effort are you willing to go to? There’s an obnoxiously obscure way to do it but it requires you to set up a dummy directory to hold surrogates for the system headers. OTOH, it doesn’t require any changes in any of your source code. The same technique works equally well for C code. … Read more
As advocated by @dulacc in his comment, __attribute__ ((optnone)) works on clang 9.0.0 on Mac’s High Sierra.
This is a C++20 feature that allows aggregate initialization through the standard constructor syntax, rather than the typical braced-list initialization syntax. (Note that this only works if the parameters cannot be used in a valid call to a default or copy/move constructor. If they could, that would be called instead of performing aggregate initialization.) According … Read more
This is a pretty deep rabbit hole, and I don’t know if I’ve explored all of its twists and turns yet. But here’s a first draft of an answer; suggestions for improvement are welcome. At its core, the culprit is the so-called “fused multiply-add” (or, in this case, a fused multiply-subtract). Fused multiply-add is a … Read more
I believe I have found the relevant piece of standardese. ยง14.8.2p7 says: The substitution occurs in all types and expressions that are used in the function type and in template parameter declarations. Since EnableIf<dependent_true_type<T>> is used in a template parameter declaration, substitution should occur and this is a bug in clang.
GCC bug. [dcl.spec.auto]/p7 (quoting N4527): When a variable declared using a placeholder type is initialized, […] the deduced return type or variable type is determined from the type of its initializer. […] Otherwise, let T be the declared type of the variable […]. If the placeholder is the auto type-specifier, the deduced type is determined … Read more
I had a response on the cfe-dev (clang front end developers’ list) from one of the developers: LLVM currently always spills “rm” constraints in order to simplify the handling of inline asm in the backend (you can ask on llvmdev if you want details). I don’t know of any plans to fix this in the … Read more
According to the standard 27.9.1.11 basic_ofstream constructors or, its more “readable” version http://en.cppreference.com/w/cpp/io/basic_ofstream/basic_ofstream , std::basic_ostream<> has a move constructor, so the code should compile. clang++ 3.5 compiles it with -std=c++11 or -std=c++1y. Also gcc5 compiles it, so probably it is not implemented in libstdc++ for gcc < 5 Interestingly, the lack of move semantics is … Read more