May 2006 - Darwin-dev - lists.apple.com

Re: StartupItems
by Garance A Drosihn 24 May '06

24 May '06

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Sounds like a reasonable position to me. -- Garance Alistair Drosehn = gad(a)gilead.netel.rpi.edu Senior Systems Programmer or gad(a)freebsd.org Rensselaer Polytechnic Institute or drosih(a)rpi.edu _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… At 10:45 PM -1000 5/22/06, Jordan K. Hubbard wrote: Launchd is but one arrow in this quiver, as are the various notification mechanisms (notify(3), configd, et al) which provide ways of finding out when things like "disk availability" have occurred WITHOUT "constantly checking" (to use your suggestion) in a performance-impacting way. I'm also sure that none of these mechanisms are close to perfect or all-encompassing, which is why we need people to file radars asking for enhancements. You're also on darwin-dev, a list which is specifically dedicated to dealing with MacOSX-related issues. We're interested in "general UNIX problems", to be sure, but a solution which only works on MacOSX is still a solution. If other UNIXen are interested in solving some or all of the sorts of problems I've talked about then we'll be more than happy to investigate their solutions as well, but please don't suggest that we're all constrained to run only as fast as the slowest runner. That's merely an argument for mediocrity and maintaining a status-quo which frankly sucks. It's 2006 already - enough with the clinging to 80's technologies and approaches which could not possibly have anticipated the world we have today. Back in the early 1990's I had a NeXTstation. I needed it to work with DHCP. At the time, many experienced Unix systems-programmers and sys-admins told me what an ignorant slob I was, and that "unix HAD to have a fixed-ip address. It just WASN'T possible to reliably setup unix with DHCP", and that I was a fool for even trying. After awhile I gave up on the "experienced" wizards who knew all and did nothing, and I figured out how to get my NeXTstation to do DHCP. And it worked fine, although admittedly it took me a few weeks to find everything which wasn't expecting the IP address to change, and figure out a solution for all of them. But in the end it took me less time to do it myself, than the time I had wasted trying to convince these experienced wizards that I needed to have it work. I had working DHCP, and I didn't have to reboot the machine every time the address changed. Hopefully all those experienced wizards are still demanding static IP addresses for their laptops. Sometimes experience is the worst enemy of progress. My guess is that if launchd is improved to cover the issues that it does not already cover, then five years from now the experienced wizards will be just as happy with the result as anyone else. And yes, I run freebsd, and we already have NetBSD's rc.d. That was a vast improvement over what we had before then, but there's nothing written in stone saying there can not possibly be an even better way to do things. It seems to me that it's worth a try. This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by André-John Mas 24 May '06

24 May '06

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Hi, Maybe there is already such a document, but I am not aware of one. Andre _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Myself I will just accept the reality of how things are. If I feel there is a better solution then I would share it with everyone, but the truth is I have other issues to deal with and feel that there enough qualified people working on launchd that I don't need to worry about the issues unless I come to road block. What I would appreciate, and maybe this may help many people, is a good FAQ on how to make programs work with launchd. I see the form: "I have service 'x', which is a type of 'z', how do recommend I make it work with launchd?". Examples of type 'z' that I can think of off the top of my head, very quickly: - servers with short start up times, suitable for "inet config". ssh for example - servers with long start up times and are expected to launch and wait. Tomcat for example - network tunnel This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by seebs＠plethora.net 24 May '06

24 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <4473B7CE.6000204(a)cse.unsw.edu.au>, Charles Gray writes: >The over-engineering possibilities are endless, but the base app seems >trivial. I'm sure there are a few sysadmins and other people with >similar problems who's like to see such an app available, too. Yes. And if it were done well it could go into the system, so everyone could use it, and it could be available for startup tasks. Which is sort of the point. Right now, launchd is the world's best replacement for inetd and cron, and doesn't even pretend to address the issues that startup systems usually address. Now, it may be right that these should be addressed elsewhere; in fact, there's a lot of arguments for it. But to say that they should be addressed elsewhere is to say, not "launchd addresses this very well by not even allowing for it", but "launchd doesn't do this, use a better tool for that task". I don't think launchd is a good match for system startup. It's not trying to provide any of the services that startup generally wants. If you want a good startup system that deals well with dependency ordering and such, get NetBSD's rc.d. It works and it solves the problem, and it doesn't step on launchd's area of expertise (on-demand launching) at all. -s _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by Charles Gray 24 May '06

24 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Debian Thunderbird 1.0.7 (X11/20051017) In message <4472D0D3.8000002(a)teaching.physics.ox.ac.uk>, Graham J Lee writes: Hmm. Well, that suggests a way to resolve this fairly elegantly; just provide a hook in plists for having services Do Something which causes other services to run. So, say, you add a tag like <key>StartupDone</key> <string>Gzornenplatzen</string> in some plists, and in others, you have <key>StartupAfter</key> <string>Gzornenplatzen</string> Bam. Nice generic way to put the "something happened and I want to be run" *IN LAUNCHD WHERE IT BELONGS* rather than having all the applications sit around inventing new notification methods and writing their own code. That's the thing; in many cases, it's utterly trivial to explain what ordering I want. If I have an ordering tool, I need one or two lines of code somewhere. If I have to do all the arbitration by hand, I can end up doing callbacks and daemon communications in two or three different protocols at once. I'm confused as to why this is such a big deal for you. Chuck _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Peter Seebach wrote: I think the argument would be that system startup is just a specific example of the general case "something interesting happened and I want to be run because of it", and that launchd deals with the happening of interesting things. I'd agree with that, if it were the argument :-) I agree both sides of the debate have a point. I think launchd has an excellent model and I don't think it should have the hooks you describe. I also agree, however, that there is a good case for support lowest-common-denominator apps. The problem is your assertion this this should be forced into launchd, where it doesn't belond. So write "an ordering tool", like you describe. It seems like it would be trivial to write a small program which waits for the launchd disk events you want and then runs a program. This could be your (unmodified) app, or someone else's apps, or a script which runs a bunch of scripts in an rc-like directory. As a bonus, thanks to launchd, in the event of a "it'll never happen" loss of volume (or just someone using your software over NFS and a flaky network), when things go away the same mechanism can request an orderly shutdown via scripts, kill or whatever. The over-engineering possibilities are endless, but the base app seems trivial. I'm sure there are a few sysadmins and other people with similar problems who's like to see such an app available, too. This email sent to site_archiver(a)lists.apple.com

1 0

gdb and mach exceptions
by Brendan Creane 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=NdiuJFn4s+lfeIit+DGT/oenZzxHWcIZ89YB0hLTO9XeYW7gxSpE3EYSfM85exTh6UGb5kfodTraUAxYyVMwWz26cfOwBzuX7lYZT7W5PaSJLpRVFpZkGt8UtfBEAKPhRFgfskuZqskjo/Hi9ScqW2sG3a5VR0EIbgcyy2R9eG8= ; Hello All, I have a simple but still unsolved problem involving the interaction of the EXC_BAD_ACCESS exception and gdb: I can't figure out how to get gdb to pass the exception to my signal handler. My test program works fine if I send a SIGUSR1 -- gdb passes the signal to my program and I get a breakpoint in the signal handler. The test program works as expected w/o gdb -- the EXC_BAD_ACCESS invokes my sigaction handler. But provoking a EXC_BAD_ACCESS within gdb stalls on the offending code. I'm curious if there is analagous gdb command "handle SIGUSR1 nostop" for EXC_BAD_ACCESS? Is there a different way from sigaction() to specify how to handle the EXC_BAD_ACCESS exception? thanks in advance for your help, cheers, Brendan Creane _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by Dan Shoop 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I joined this list looking for the latest info on issues (bugs) that other OS X developers were finding so I could work more efficently. It costs me about the same effort to write 1 line of code, or write 10 lines of prose, or read 100 lines. I've spent way too much of my limited time scanning all this launchd debate and emptying my email inbox. At this rate, reading all this dogma and reminiscent dialog about how things should be is less effective than blindly coding around any OS related problems I encounter. Am I on the wrong list? Where do the real programmers (with tight schedules like mine) exchange useful info? irc? -- -dhan ------------------------------------------------------------------------ Dan Shoop AIM: iWiring Systems & Networks Architect http://www.ustsvs.com/ shoop(a)iwiring.net http://www.iwiring.net/ 1-714-363-1174 pgp key fingerprint: FAC0 9434 B5A5 24A8 D0AF 12B1 7840 3BE7 3736 DE0B iWiring provides systems and networks support for Mac OS X, unix, and Open Source application technologies at affordable rates. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… At 11:24 AM -0700 5/23/06, JR wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by Kevin Van Vechten 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 23, 2006, at 2:23 AM, Peter Seebach wrote: So, say, you add a tag like <key>StartupDone</key> <string>Gzornenplatzen</string> in some plists, and in others, you have <key>StartupAfter</key> <string>Gzornenplatzen</string> - Kevin _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Well, that suggests a way to resolve this fairly elegantly; just provide a hook in plists for having services Do Something which causes other services to run. Bam. Nice generic way to put the "something happened and I want to be run" *IN LAUNCHD WHERE IT BELONGS* rather than having all the applications sit around inventing new notification methods and writing their own code. Sure, but the issue you're overlooking is that there's no universal definition for when startup is done. Disk and network availability are events, like keystrokes. When do you know that keyboard input is done? You could possibly wait for the keyboard to be quiescent. You could alternatively look for a marker like newline. But you'll never get an EOF. While I agree with you that sample code / libraries would be help applications approximate when startup is done (just like Libc provides line-buffered stdin), I disagree that this belongs in launchd itself (or keyboard device drivers). Launchd triggers are unambiguous and discrete, there isn't really room for interpretation. Adding in keys like "StartupDone" changes that. This email sent to site_archiver(a)lists.apple.com

1 0

OS X developer lists (was Re: StartupItems)
by Peter Bierman 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I joined this list looking for the latest info on issues (bugs) that other OS X developers were finding so I could work more efficently. It costs me about the same effort to write 1 line of code, or write 10 lines of prose, or read 100 lines. I've spent way too much of my limited time scanning all this launchd debate and emptying my email inbox. At this rate, reading all this dogma and reminiscent dialog about how things should be is less effective than blindly coding around any OS related problems I encounter. Am I on the wrong list? Where do the real programmers (with tight schedules like mine) exchange useful info? Here. It's just that Darwin has no bugs, so we had nothing else to talk about. ;-) http://www.omnigroup.com/mailman/listinfo/macosx-dev -pmb _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… At 11:24 AM -0700 5/23/06, JR wrote: Seriously, if you want to read about a variety of Mac OS X (not just Darwin) programming issues, join the OmniGroup macosx-dev list. This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by JR 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I joined this list looking for the latest info on issues (bugs) that other OS X developers were finding so I could work more efficently. It costs me about the same effort to write 1 line of code, or write 10 lines of prose, or read 100 lines. I've spent way too much of my limited time scanning all this launchd debate and emptying my email inbox. At this rate, reading all this dogma and reminiscent dialog about how things should be is less effective than blindly coding around any OS related problems I encounter. Am I on the wrong list? Where do the real programmers (with tight schedules like mine) exchange useful info? JR _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: StartupItems
by Chris Ridd 23 May '06

23 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AcZ+lIPpwjxiO+qHEdqzKwARJHlZbA== Thread-topic: StartupItems User-agent: Microsoft-Entourage/11.2.3.060209 On 23/5/06 7:03, Dan Shoop <shoop(a)iwiring.net> wrote: > And if you think that the rc.d mechanism used so commonly today is an > improvement or so great try answering the following common question: > "what letter/number/name should I make my rc file to start up X? > Which rc directory should it go under?" Such mechanisms are a huge > step backwards and unbelievably fragile. It is interesting that Sun also came to the same conclusion, and replaced it with their Service Management Framework (SMF) in Solaris 10. Has anyone compared launchd with SMF? Is there anything SMF does better than launchd? Cheers, Chris _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0