On Jan 5, 2008 5:25 PM, Eric Blake <[EMAIL PROTECTED]> wrote: > Ultimately, if this is a real issue in practice, the solution would be a > gnulib module that provides waitpid replacements for lacking systems. But > most likely, nobody actively ports to those old platforms any more. Also, > gnulib already has a wait-process module, so I wonder if it would be > useful here.
I looked at it, and it's not. The main problem is that too much information about the fate of the process is irreversibly (afaict) conflated into a single return value. The xargs program needs to do different things depending on whether the program exited or was killed by a fatal signal, for example. When we fix the issue with child processes that exit with status 126, this will get a little more complex, too (xargs should return 123 for that case, but actually returns 126, iirc). James.
