When closing descriptors, set some fixed (configurable?) cap about daemonize

[miigotu]

Is disabling closing of file descriptors altogether and managing your pre-daemonization descriptors yourself a proper way to work around this?

[devgs]

@miigotu As an option - definitely. But i would suggest adding an optional function parameter (callback) that should handle the closing, to override the built-in logic.

Moreover, the closing should definitely be handled by the child side of the post-fork pair. Imagine a scenario, when the pipe is created. One side (descriptor) belongs to a parent and the other side to the child (now daemon). Parent process can't close any of them before doing the fork. It's the child's responsibility to close the parent's side and vice versa. But both must do it post-fork.