— 📝Fixes#1315, #1374, and #1400.
— 📝 Lifetime in iterators was referencing the wrong stack (the main thread) rather than the coroutine's stack at time of creation.
— 📝 Using main_reference/main_* objects was a suitable enough fix for most of these problems.
— ⚡🛠 Prevent performance and usability issues from changing containers by storing the being/end iterator separately, rather than continually invoking `deferred_uc::end(…)` every time.
— 🛠 Improve sizes for stored iterators in select cases.
— 🛠 Allow for sentinel-style C++20-and-beyond ranges.
— 🔧 Improve single file generation CMake.
— 👷♀️ Fix up internal Lua build system issues.
[why]
Compiling with clang issues this warning (clang 10.0 and 15.0 tested):
.../sol/sol.hpp:14541:32: warning: unknown warning group '-Wmaybe-uninitialized', ignored [-Wunknown-warning-option]
#pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
[how]
Check if the compiler does know the warning, before trying to turn the
warning off.
Signed-off-by: Fini Jastrow <ulf.fini.jastrow@desy.de>
— Lua 5.1, 5.2, and 5.3 seemed to have a deficiency in which the stack thread space variable (given as a lua_State*) would die before everything referencing it would be properly dead. This made holding a reference for keep-alive purposes impossible to maintain. Therefore, we retrieve the main thread to keep it alive.
— 🛠 Update SOL_IS_(DEFAULT_)ON/OFF usage to be more idiomatic and less confusing (add SOL_RAW_* alternatives as well)
— 💚 Re-check CI
— 👷♀️ Add missing header from ebco.hpp
- Add a flag to allow size_t(-1) and similar shenanigans (SOL_ALL_INTEGER_VALUES_FIT)
- Half-fix, but not fully, for #1183, #1072, #1038, #965
- Fix#1126
- Prepare for #1061
- It's not at all safe. And I do not at all care - table iterators are inefficient and terrible, and there's nothing we can do to make this situation even remotely better.
