Re: StartupItems
Re: StartupItems
- Subject: Re: StartupItems
- From: Kevin Van Vechten <email@hidden>
- Date: Mon, 22 May 2006 18:47:54 -0700
Andre, I think you're right on the mark, conceptually... however I
think it's really more sample code that's necessary, not a GUI
utility. (There's already a GUI utility you might want to check out
<http://lingon.sourceforge.net/>).
Sample code that can perform the tasks in question, such as "wait
until the network interfaces seem to be up" and "wait until the disks
seem to have mounted," would probably go a long way toward
facilitating the adoption of launchd. (It's really not that much
code to accomplish these tasks).
Michael's correct that sysadmins should not need to worry about
this. I think the burden is on Apple to provide the necessary
information, and on developers (commercial or open source) to adopt
the technology. Will it happen overnight? Certainly not. You can
see that not even all of Apple's StartupItems were converted to
launchd plists for Tiger. But launchd is the direction things are
moving, and I think the benefits will become more apparent as we draw
nearer the destination.
Also, for what it's worth, there's a WWDC 2006 session scheduled:
"Daemon Wrangling with Launchd."
- Kevin
On May 22, 2006, at 6:15 PM, André-John Mas wrote:
Probably a suitable compromise would be come up with GUI based
utility that facilitates the creation of launchd based services. I
am not sure what sort of form it would take on, but sometimes the
best alternative is provide a suitable tool that facilitates the
use of the technology at hand.
Andre
On 22-May-06, at 18:00 , Michael Bartosh wrote:
On May 22, 2006, at 10:40 PM, Kevin Van Vechten wrote:
I'd be surprised if there is not a platform specific line of code
in your program. You can wrap your launchd support in #ifdef
__LAUNCHD__ or something equivalent. It's overstating the
problem to say that you can't have a unified code base. Adding
Launchd support is a relatively small amount of code, I know from
experience. I think your sysadmins (customers) will ultimately
be pleasantly surprised by this investment. Your server can be
automatically kept alive. This and other servers can be
uniformly launched, stopped, and queried via launchctl or
programmatically.
That's fine. But what about the poor, dumb sysadmin who -can't
write code-. These guys are not developers. Most sysadmins aren't.
They take code developed by other people and deploy it themselves.
And they need consistent ordering and dependencies that they can
impose on that code.
You (Kevin) seem to acknowledge this when someone brings it up.
But the messaging from Apple seems to be 'use launchd, it's the
wave of the future!'
not
'You might want to use launchd eventually, it has some cool
features, but will create headaches for you in the short run'.
When the latter is far more accurate.
You're right. Launchd is cool. It does some very cool things. 90%
of the time those things are not accessible without dependency
checking various systems require.
I mean, at least add something to launchd.plist that can talk to
configd.. wait for specific configd states, like configd's kicker
does. That alone would make launchd -a lot- more useful.
-mb
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Darwin-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden