Re: Library support for two-phase daemonization

[Anthony Mallet <anthony.mallet%laas.fr@localhost>]

On Wednesday, at 19:29, Christos Zoulas wrote:
| In article <21224.54743.92353.870553%guava.gson.org@localhost>,
| >Implementation nits aside, do you find the API acceptable, or would
| >you prefer to expose the file descriptor as Joerg is suggesting?
| 
| I think that exposing the file descriptor is nice because it avoids the
| static storage issue.

But this makes the API more cumbersome.

What about a sem_open(3) for the synchronization, initially empty and
sem_post'ed by the child?
The semantics may be clearer, but sadly this requires -lrt.

Or maybe just a kill(SIGHUP) to the parent?