Closed mucki-at closed 4 years ago
Thanks for the report. Based on your patch, I've implemented a solution that also works in C++03 and ensures static initialization based on existing boost singleton implementations.
Thank you, much appreciated. Unfortunately it doesn't quite fix the issue (see my comments in your commit)
On Sat, May 23, 2020 at 2:40 PM Ion Gaztañaga notifications@github.com wrote:
Closed #148 https://github.com/boostorg/container/pull/148 via 093467e https://github.com/boostorg/container/commit/093467e1d3e05b93b48e38517750239c6649a63e .
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/boostorg/container/pull/148#event-3367446266, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFR6C7IBVYCZJWJRE6L5ZDDRTA7FJANCNFSM4NFO2FLA .
Let's see if this fixes the issue and also supports C++03
make sure that pmr global resources are constructed on first use so that they can be successfully used during static initialization (ie: before main()). Caveats: unless "magic statics" are supported by the compiler the very first call to new_delete_resource() and null_memory_resource() is not thread-safe.