Re: Using glib's g_file_monitor_file and g_file_monitor_directory

[Paul Eggert]

On 05/28/2013 03:39 AM, Michael Albinus wrote:
> Comments on the patch are welcome. And we should also start to design an
> upper layer over gfilenotify.c, inotify.c and w32notify.c, which
> abstracts from the respective file notification events

I've been running into a similar issue with monitoring
child processes; see Bug#14474.  A couple of dumb questions:

Should the issues of file-change and process-change notifications
be combined, or kept separate?

Would it make sense, at the low level, for Emacs to rely on the
glib primitives in child-process area?  I worry about race
conditions (SIGCHLD arriving in the middle of memory allocation,
that sort of thing).