Connor Fitzgerald
Connor Fitzgerald
I license past and future contributions under the triple MIT OR Apache-2.0 OR Zlib license, allowing licensees to choose any one at their option. Glad to see it getting more...
The standard solution is to use `std::pmr::vector` etc. which are just aliases for `std::vector` and so on. It truly is the solution to the C++ allocator nonsense. The `std::pmr::polymorphic_allocator` accepts...
Good stuff, aiming for the allocator method? --- Connor Fitzgerald Sent from my Phone. Please excuse my brevity. On November 17, 2018 9:40:48 AM EST, RichieSams wrote: >I'm working on...
Take a look at Google's abseil library and how they implement std::any, std::variant, etc. --- Connor Fitzgerald Sent from my Phone. Please excuse my brevity. On November 17, 2018 11:48:29...
Let me check, haven't tried it.
Yeah doesn't seem to help. Still end up running out of fiber slots.
This appears to be a race condition within the thread local storage of the fibers. There are a couple of race conditions TSan has caught, but some of them may...
I think we should be able to get away with just `std::memory_order_acquire` as that's a barrier for code from below rising above. Note that this is with optimization off and...
Raising it all the way to `std::memory_order_seq_cst` didn't do anything, we still have the same problem.
To better help me understand what is going on, what does that call force synchronization of that allows the delete _and_ erase to fire?