site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com User-agent: Thunderbird 1.5.0.4 (Macintosh/20060530) On Jul 8, 2006, at 12:45 PM, Nicola Vitacolonna wrote: Cheers, Graham. -- Graham Lee UNIX Systems Manager, Oxford Physics Practical Course http://iamleeg.blogspot.com/ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... On 10/7/06 18:16, Dave Zarzycki wrote: 2) Is launchd to be preferred over other methods in this case? And is it the way to go in general? The preferred solution does not involve launchctl or programmatic control whatsoever. In Apple's idealized view of the universe, your VPN daemon would be launch on demand via some IPC based mechanism. This has the nice side effect of implicit correctness. From your client's perspective, the daemon is virtually running all the time. Purely a mental exercise at the moment, but in the Apple universe what's the approach for daemons which service IPC but also have a long startup time? For instance, say I create a leegsql server which, on a typical box, spends "a while" checking and precaching the tables when it loads up. This while is long enough that users would notice if it were started and stopped on demand. Would the "expected" approach be to launch the whole thing at startup, or to launch the sqld at startup and have on-demand service daemons transiently created when network, ODBC, JDBC etc requests appear? This email sent to site_archiver@lists.apple.com