November 2007 - Darwin-dev - lists.apple.com

Re: 64-bit srandom()?
by Michael Smith 15 Nov '07

15 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Nov 14, 2007, at 12:04 PM, darwin-dev-request(a)lists.apple.com wrote: int32_t seed=(int32_t)(time(NULL) & 0xff); srandom(seed); That seems more like the least-significant 8 bits. = Mike _______________________________________________ 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… Note that strictly speaking, seeding srandom with time may not be exactly secure, but then, srandom's not a cryptographically secure RNG anyway. More to the point, the *only* time you want to use srandom() is when you want a reproducible set of pseudo-random numbers. If your original objective was to obtain random numbers, you should be reading bytes from /dev/random or similar. This email sent to site_archiver(a)lists.apple.com

1 0

Re: 64-bit srandom()?
by Clark Cox 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=IKD+bTi3jMyQ0jgDSlQWohroY6QR6r6xXv6Kvc200IQ=; b=uqpvQX8H9nIgdpP1pz8jVofXVNWnQ0TocdWGijqDJFn3QBe0piNNQHND1Ty+dT+bVAJUWc3jU3FxE1pYQFXsL9q9yKDPK/h7Tin5BkUCRlj1Gu+hu0MVVYCtN3r3j8VYrcm+4Av2xi9GuaymJBmv9i/Wgoqi2Ct5WkuM92QruGk= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=Gd6JMOvNF2uA5B4lGTXX3BUbqRRPQYNAOTYnZ+HZlX4rBqfBt/sUg2LsTyroPmblgA8IYgVtWPiJ1f2xVs4oWcE99yZ2sx0LAgK9mBu/ddWLAGYgijCXG8d+LxiwFkIQujBhxxOl2G5JXGmiuZOUgs+hRzQ3dAUZ6yKurGl8qiE= On Nov 14, 2007 10:57 AM, A.M. <agentm(a)themactionfaction.com> wrote: > > On Nov 14, 2007, at 1:37 PM, Nick Zitzmann wrote: > > > I searched around and didn't see anything relevant... > > > > In our 32-bit projects, we're calling: > > > > srandom(time(NULL)); > > > > What's the best way of doing something like the above in a 64-bit > > program? srandom() takes a 32-bit integer when compiling for both > > 32-bit and 64-bit, but time() returns a 64-bit integer when > > compiling for 64-bit. I guess I have until 2038 to worry about > > this, but I'd rather deal with it now... > > srandrom() doesn't really care about the actual time- you should just > make sure you seed with the least-significant 32-bits which you can Because of the type conversion rules of C, you're guaranteed to get the least significant 32-bits when you convert a 64-bit unsigned integer to a 32-bit integer. No need for any bitwise operations.: unsigned seed = time(NULL); -- Clark S. Cox III clarkcox3(a)gmail.com _______________________________________________ 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: 64-bit srandom()?
by seebs＠seebs.net 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <9A1D3A66-FE05-4F2B-BDCC-58A1D829339A(a)themactionfaction.com>, "A.M." writes: >srandrom() doesn't really care about the actual time- you should just >make sure you seed with the least-significant 32-bits Exactly. >int32_t seed=(int32_t)(time(NULL) & 0xff); >srandom(seed); That seems more like the least-significant 8 bits. Note that strictly speaking, seeding srandom with time may not be exactly secure, but then, srandom's not a cryptographically secure RNG anyway. -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: 64-bit srandom()?
by A.M. 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Nov 14, 2007, at 1:37 PM, Nick Zitzmann wrote: I searched around and didn't see anything relevant... In our 32-bit projects, we're calling: srandom(time(NULL)); int32_t seed=(int32_t)(time(NULL) & 0xff); srandom(seed); (Or something like that- untested code.) Cheers, M _______________________________________________ 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… What's the best way of doing something like the above in a 64-bit program? srandom() takes a 32-bit integer when compiling for both 32-bit and 64-bit, but time() returns a 64-bit integer when compiling for 64-bit. I guess I have until 2038 to worry about this, but I'd rather deal with it now... srandrom() doesn't really care about the actual time- you should just make sure you seed with the least-significant 32-bits which you can do with: This email sent to site_archiver(a)lists.apple.com

1 0

Re: 64-bit srandom()?
by seebs＠seebs.net 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <902354B2-40D4-4ECC-8307-39696A3C4266(a)chronosnet.com>, Nick Zitzmann writes: >I searched around and didn't see anything relevant... > >In our 32-bit projects, we're calling: > >srandom(time(NULL)); > >What's the best way of doing something like the above in a 64-bit >program? srandom() takes a 32-bit integer when compiling for both 32- >bit and 64-bit, but time() returns a 64-bit integer when compiling for >64-bit. I guess I have until 2038 to worry about this, but I'd rather >deal with it now... Why would you change it? The integer will get politely converted down (it's an unsigned long, so overflow's safely handled) and it'll do what you want. -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

64-bit srandom()?
by Nick Zitzmann 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I searched around and didn't see anything relevant... In our 32-bit projects, we're calling: srandom(time(NULL)); Nick Zitzmann <http://www.chronosnet.com/> _______________________________________________ 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… What's the best way of doing something like the above in a 64-bit program? srandom() takes a 32-bit integer when compiling for both 32- bit and 64-bit, but time() returns a 64-bit integer when compiling for 64-bit. I guess I have until 2038 to worry about this, but I'd rather deal with it now... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Launchd Agents, Login Items, and Tiger
by seebs＠seebs.net 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <358F23D4-61D6-44E5-B91D-4B7986377421(a)alum.mit.edu>, Mike Kobb write s: >Okay, thanks much! It does seem obvious now that I see it! :-) I can >keep track of how quickly the app quits so that I can abort if it >seems to be dying on launch, and I guess I'd just have the script >check the return value of the program for a "normal" result to >terminate the loop if I need to actually shut down the process. Sure. TOOFAST=5 x=$(date +%s) until MyProgram do y=$(date +%s) if [ $(expr $y - $x) -lt $TOOFAST ] then exit 1 fi x=$y done ... And this is why I like launchd, because I've written WAY too many copies of that program. -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: Launchd Agents, Login Items, and Tiger
by Mike Kobb 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com --Mike On Nov 14, 2007, at 8:29 AM, Dave Zarzycki wrote: On Nov 13, 2007, at 6:25 PM, Peter Seebach wrote: Hmmm. You might be able to do something clever. If using launchd won't cut it, is there an alternative approach for not just launching me at login, but also re-launching me if I crash? #!/bin/sh while true do MyProgram done Obviously, I don't anticipate crashing, but the project requirements I've been handed specify automatic re-launching of this background helper app, and I'd prefer to find a way to satisfy the requirements rather than argue to have the requirement removed. I can think of other alternatives, but they are messier. _______________________________________________ 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… Okay, thanks much! It does seem obvious now that I see it! :-) I can keep track of how quickly the app quits so that I can abort if it seems to be dying on launch, and I guess I'd just have the script check the return value of the program for a "normal" result to terminate the loop if I need to actually shut down the process. Agreed. It makes sense -- but I don't know of a good way to do it. I think Peter's approach is the best option for Tiger automatic relaunch of programs within the session. davez This email sent to site_archiver(a)lists.apple.com

1 0

Re: Launchd Agents, Login Items, and Tiger
by Dave Zarzycki 14 Nov '07

14 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Nov 13, 2007, at 6:25 PM, Peter Seebach wrote: Hmmm. You might be able to do something clever. If using launchd won't cut it, is there an alternative approach for not just launching me at login, but also re-launching me if I crash? #!/bin/sh while true do MyProgram done Obviously, I don't anticipate crashing, but the project requirements I've been handed specify automatic re-launching of this background helper app, and I'd prefer to find a way to satisfy the requirements rather than argue to have the requirement removed. I can think of other alternatives, but they are messier. davez _______________________________________________ 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… Agreed. It makes sense -- but I don't know of a good way to do it. I think Peter's approach is the best option for Tiger automatic relaunch of programs within the session. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Launchd Agents, Login Items, and Tiger
by seebs＠seebs.net 13 Nov '07

13 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <59FAD51F-A47A-4912-943C-DAB03029F373(a)alum.mit.edu>, Mike Kobb write s: >I'm trying to work around the documented limitation in launchd (on >Tiger) that prevents it from being used to launch a process at user >login. The limitation, as I understand it, is that there's no way to >prevent launchd from launching the process even if the user logs in >via a non-GUI method like ssh. Ahh! >So, I guess what I'm asking is if, *once I have been launched as a >Login Item*, I can leverage launchd to simply re-launch me properly if >I should crash. Or, does a subsequent launch-on-crash suffer the same >basic problem that prevents launching me at login in the first place? Hmmm. You might be able to do something clever. >If using launchd won't cut it, is there an alternative approach for >not just launching me at login, but also re-launching me if I crash? #!/bin/sh while true do MyProgram done >Obviously, I don't anticipate crashing, but the project requirements >I've been handed specify automatic re-launching of this background >helper app, and I'd prefer to find a way to satisfy the requirements >rather than argue to have the requirement removed. Agreed. It makes sense -- but I don't know of a good way to do it. >PS: "KeepAlive" is the 10.5 option. In 10.4, it was "OnDemand", which >is now deprecated. It is very sad that I should have gotten this backwards. -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