(mailed landry@, but forgot to cc ports) On Tue, 23 May 2017 20:53:12 +0200 Landry Breuil <lan...@openbsd.org> wrote:
> Better with an update that actually compiles. works here, but doesnt > directly fix my own issue... strange. Worked after forcing an update of > the problematic dir via gmpc, and a rescan. > > Either way, we should go for the update if there are no regressions. > David ? Does the update fix the database issue for you? It does not seem to have any effect here, not without the #define gcc_pure bit. (0.20.8 was the first thing I checked yesterday, hoping that a simple update would be enough :) terminate called after throwing an instance of 'std::system_error' what(): Failed to access /storage/music/test: No such file or directory May 23 15:18 : exception: Failed to access /storage/music/dts/test: No such file or directory When I add the Compiler.h patch to 0.20.8, it starts working, so perhaps that commit you mentioned was not enough. Also, you have + static +-#if GCC_OLDER_THAN(5,0) ++#if GCC_OLDER_THAN(5,0) || defined(__clang__) + /* gcc 4.x has no "constexpr" for std::max() */ + const but with #define GCC_OLDER_THAN(major, minor) \ (GCC_VERSION > 0 && CLANG_VERSION == 0 && \ GCC_VERSION < GCC_MAKE_VERSION(major, minor, 0)) we end up with "if not clang or if clang", which works for us, but does not feel right. I got a bit lost in all the compiler checks defined in Compiler.h, not sure what the right way is though. (The logic seems reversed to begin with and clang 4 is supposed to support constexpr for std::max() I think.) But this compiler/library/c++ maze is way above my abilities, sorry.