Anyway, why would anybody standardize this at all? Too many ways to do it and each one sucks because of race conditions. Any program that needs such a feature s

Anyway, why would anybody standardize this at all? Too many ways to do it and each one sucks because of race conditions. Any program that needs such a feature should just write the 10-20 lines version that it deems best. There will hardly be any project that needs to know the "recursive file size" that is also too small to be able to afford this extra work. Ok, but maybe I'm missing something here: Why parallel? Is there any benefit?