In http://reviews.llvm.org/D8253#138984, @jroelofs wrote:
> Please upload your diffs with more context (i.e. `git diff -U999`)... it'll > make the reviews a bit easier. Sure! Will do that as of now. Still trying to get used to Phabricator. > How about defining `_LIBCPP_HAS_NO_REALTIME_CLOCK`, similarly to the existing > one for monotonic clocks in `include/__config` under the appropriate guards? > (i.e. it would be defined on `__APPLE__`, and anywhere else that doesn't have > `clock_gettime(CLOCK_REALTIME, x)`) So it is important to keep in mind that `_LIBCPP_HAS_NO_MONOTONIC_CLOCK` does not indicate whether `CLOCK_MONOTONIC` is present or not. It is used to indicate that the system provides some method of obtaining a monotonic clock. The `_LIBCPP_HAS_NO_MONOTONIC_CLOCK` flag is not set on OS X, even though there is a monotonic clock. Using that logic `_LIBCPP_HAS_NO_REALTIME_CLOCK` would imply that the system has no real-time clock at all. ================ Comment at: src/chrono.cpp:72 @@ -53,2 +71,3 @@ -#ifdef __APPLE__ +#ifdef CLOCK_MONOTONIC + ---------------- jroelofs wrote: > This is already taken care of where `_LIBCPP_HAS_NO_MONOTONIC_CLOCK` is > defined. I don't think there's a need to change anything around the > steady_clock implementation. The intent of this change was to be a bit more consistent with the change to system_clock. First try to use the POSIX API (`clock_gettime`) and only if that one is unavailable fall back to custom APIs specific to a smaller set of operating systems. http://reviews.llvm.org/D8253 EMAIL PREFERENCES http://reviews.llvm.org/settings/panel/emailpreferences/ _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
