* Add NetBSD Support
Includes:
- Hints to find non-compatibility libc functions
- Struct and constant definitions for sockets, polling, etc.
- Changes to the linker code to work around some quirks in the NetBSD dynamic linker
- A target triple for netbsd aarch64 so llvm builds/links the compiler properly on this platform
* Updated releasenotes and some compacting
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
* add join for ThreadPool without destroying the threads
* Make the main Thread block waiting for the worker threads to finish instead of buzy looping and do proper initialization and freeing of all variables.
* Updated test to use `atomic_store` and take into account the maximum queue size of the threadpool.
* - Add `ThreadPool` join function to wait for all threads to finish in the pool without destroying the threads.
- Return of Thread/Mutex/CondVar `destroy()` is now "@maydiscard" and should be ignored. It will return void in 0.8.0.
- Return of Mutex `unlock()` and `lock()` is now "@maydiscard" and should be ignored. They will return void in 0.8.0.
- Return of ConditionVariable `signal()` `broadcast()` and `wait()` are now "@maydiscard". They will return void in 0.8.0.
- Return of Thread `detatch()` is now "@maydiscard". It will return void in 0.8.0.
- Buffered/UnbufferedChannel, and both ThreadPools have `@maydiscard` on a set of functions. They will retunr void in 0.8.0.
- Pthread bindings correctly return Errno instead of CInt.
- Return of Thread `join()` is now "@maydiscard".
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
* Added mutex tests. Add errorcheck in safe mode for Posix threads. Make non-recursive locks fail when used recursively on Windows. Fix thread pool tests. Simple locking count.
---------
Co-authored-by: Christoffer Lerno <christoffer@aegik.com>
0.5.5 Disallow multiple `_` in a row in digits, e.g. `1__000`. #1138. Fixed toposort example. Struct/union members now correctly rejects members without storage size #1147. `math::pow` will now correctly promote integer arguments. `math::pow` will now correctly promote integer arguments. Added `new_aligned` and `alloc_aligned` functions to prevent accidental under-alignment when allocating simd. Pointer difference would fail where alignment != size (structs etc) #1150. Add test that overalignment actually works for lists. Fixed array calculation for npot2 vectors. Use native aligned alloc on Windows and POSIX. Deprecates "offset". Simplification of the Allocator interface.
