refactor: prevent extra move for get-methods with nrvo

[jirol9xa]

Motivation https://godbolt.org/z/jTd99xzf1

[DeveloperPaul123]

Thanks for the PR. If you could please address the build issues, that would be great.

[jirol9xa]

sorry, i am not able to compile it locally

[DeveloperPaul123]

Ok looks like some tests are failing. I'll have to investigate a bit more if you can't build it locally.

[jirol9xa]

I tried it on my home ubuntu 22.04 with gcc-11. Everything is fine, it is not possible to reproduce the crash.

[DeveloperPaul123]

@jirol9xa Unfortunately, this is the nature of something as low level as a thread pool. Even if it's no locally reproducible, it is still likely a bug somewhere.

Were you able to run it with thread sanitizer enabled and/or asan?

[DeveloperPaul123]

I'll try this out later on on PopOS and Ubuntu 24.04 to see if it's reproducible.

[jirol9xa]

Tried to run with tsan, and had an error even if running from trunk =)) Pretty sure I am doing something wrong

[DeveloperPaul123]

Oh no, there might be a bug then. What error do you see?

[jirol9xa]

CMake Error at /home/voffk4/fast-thread-pool/build/_deps/doctest-src/scripts/cmake/doctestAddTests.cmake:45 (message):
Error running test executable
'thread-pool/build/test/thread-pool-tests':

Result: 66
Output: 

[DeveloperPaul123]

CMake Error at /home/voffk4/fast-thread-pool/build/_deps/doctest-src/scripts/cmake/doctestAddTests.cmake:45 (message): Error running test executable 'thread-pool/build/test/thread-pool-tests':

Result: 66
Output: 

I was able to reproduce this. https://stackoverflow.com/questions/77850769/fatal-threadsanitizer-unexpected-memory-mapping-when-running-on-linux-kernels this stackoverflow helped me find a workaround. Setting sudo sysctl vm.mmap_rnd_bits=28 fixed the build issue for me, but I think TSan is giving false positives. I'll have to dig in a bit more. When I run the tests without TSan they all pass.

[DeveloperPaul123]

Welp, I re-ran the test and it passed this time 🤷‍♂️