July 2005 - Darwin-dev - lists.apple.com

Re: build hybrid binary
by Alex v. Below 29 Jul '05

29 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 29.07.2005 um 09:30 schrieb plumber: how to build hybrid unix binary ? where i can find an official SDK and documentation all i found is "blabla" gcc -arch option is your friend. Alex _______________________________________________ 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… I don't know what "hybrid binaries" are, but if you mean Universal binaries (or "Fat", as the old timers call them), then the This email sent to site_archiver(a)lists.apple.com

1 0

build hybrid binary
by plumber 29 Jul '05

29 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com how to build hybrid unix binary ? where i can find an official SDK and documentation all i found is "blabla" Best Regards -----BEGIN PLUMBER CODE BLOCK----- Version: 1.0 ¬ | | > Regards Plumber || GNU Darwin. ¬ \\\ plumber.gnu-darwin.org ¬ ¬ | | > OOP --\///\ (0)=(0) Darwin/Power Mac G4. ¬ | | > ( __ò ó__ ) The best for Fede ¬ \\\ federicafontana.it ¬ ------END PLUMBER CODE BLOCK------ -----BEGIN GEEK CODE BLOCK----- Version: 3.1 GMU/S d+@ s: a- K+++++ PGP++++ K---- dpu s--:-- C++++ C-- B L X U++++ P+++ C-- P! L--- E+++ W+++ P! L--- N+++ o-- K+++++ ------END GEEK CODE BLOCK------ Plain, Loving, Useful, Mischievous, Brutal, Explosive, Responsible _______________________________________________ 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… I"It's making my eyes better. And it's legal. I could walk up to the President and blow smoke in his stupid monkey face, and he'd just have to sit there groovin' on it!" Homer J Simpson, Springfield, Illinois, Massachusetts nobody knows :) This email sent to site_archiver(a)lists.apple.com

1 0

Re: GCC
by Sam Hart 28 Jul '05

28 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com John wrote: What are the different gcc modules? What's the diff between gcc, gcc_os, gcc_os_35, and gccfast? they are used for building different parts of the system gcc-4061 is gcc 4.0 gcc_os-1809 is gcc 3.3 gccfast-1621.1 is also 3.3 ( but faster ) gcc_os_35-3506 is gcc 3.5 remember that these are all from apple's gcc branch gccfast ? or possibly gcc 4 I guess it also depends on compile options gcc-4061 ? what does that mean ? gcc_legacy used to be 2.95.2 I cannot see it in the Tiger source but it was there in Panther as long as they are needed depends on what you want to do with it S a 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… I seem to remember that most the system builds with 3.3 but I could be wrong Which one is gcc 3.1, 3.3, and 4.0? Which is the fastest? Which is the most recent? of course you could always check out the latest version of apple's branch at fsf ( YMMV ) Which is the most correct? What happened to gcc 2.9? How long will each be supported? Lastly, for someone building their own, which one is the best choice? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Why does free() take 1,100 microseconds?
by Jan Brittenson 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mozilla Thunderbird 1.0.2 (Macintosh/20050317) Markus Hitter wrote: That wasn't being discussed. _______________________________________________ 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… Especially, if you handle some 500 MB of data. Totally inappropriate to wire down such amounts. This email sent to site_archiver(a)lists.apple.com

1 0

Re: GCC
by Kevin Van Vechten 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com For Panther, 'gcc' is gcc-3.3 and 'gcc_os' is gcc-3.1. For Jaguar, 'gcc' is gcc-2.95.2 and 'gcc3' is gcc-3.1. For Puma and Cheetah there was only gcc-2.95.2. - Kevin On Jul 26, 2005, at 10:28 PM, John Davidorff Pell wrote: Lastly, for someone building their own, which one is the best choice? Thanx, JP _______________________________________________ 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… The 'gcc' project is the source for the default compiler. The 'gcc_os' project is the source for the compiler that compiles the operating system. For Tiger, 'gcc' is gcc-4.0, and 'gcc_os' is gcc-3.3. 'gcc_os_35' is gcc-3.5 and was used to compile a very limited set of system components that have ppc64 support. gcc-3.5 did not ship with Tiger, and ppc64 apps that you compile should be compiled with gcc-4.0. There's no "correct" one to use in all circumstances, but the general rule is that you should use the default compiler for a system, and only switch back to an older version if your application would not otherwise be backward compatible with a previous OS release. If you're compiling Darwin projects that you want to use to replace existing OS components, use the gcc_os compiler. I can't speak for how long any of these will be supported, but the trend from the releases mentioned above is that each release contains a new default compiler, and the default compiler from the previous release becomes gcc_os. The recent question about building gccfast reminded me of some confusion I've had recently. What are the different gcc modules? What's the diff between gcc, gcc_os, gcc_os_35, and gccfast? Which is the fastest? Which is the most recent? Which is the most correct? What happened to gcc 2.9? Which one is gcc 3.1, 3.3, and 4.0? How long will each be supported? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Why does free() take 1,100 microseconds?
by Markus Hitter 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 27.07.2005 um 16:22 schrieb John Francini: Cheers, Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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… The earlier lines of discussion made it obvious that this application needs to have near-real-time levels of predictable response. Having the OS decide to page on you when you really don't want it to do so doesn't help, and makes the platform look like it's inhospitable to real-time use. Darwin is quite capable of handling near realtime demands, even without hackery. The OS pages when there is insufficient physical memory, only, of course. If it does, it tries to page out stuff which wasn't used for the longest time. So, if your app data gets paged, it either wasn't used/touched for a long time or memory is really tight. In both situations you shouldn't modify the default behaviour. If you want to make fairly sure some portion of data isn't the next candidate to be swapped, "touch" it by reading a few bytes from each page. Only touch data you expect to be used soon, of course. Especially, if you handle some 500 MB of data. Totally inappropriate to wire down such amounts. This email sent to site_archiver(a)lists.apple.com

1 0

Re: os x virtual device
by Finlay Dobbie 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=cZVOTVVKqAMk2iGp/ewtNM/rSaE0uC7D9q50GWiEs9GDpdKvhXdL15sabQ/8HenzQSlHwfC3kceZhDt+5B+ck4WENlUKcKgLOIS8ybwH1Rpib2R0MEw6DaSuV5fE/fcCi/hncOIlLN2QRn8wEU9w0bWNgXN+5StsGChPTrHABf0= On 7/27/05, Georges Engelmann <admin(a)xliners.org> wrote: > Hello list, > > I want to write an application which provides a "virtual volume" to > mac os x. > > is this somehow possible? Of course. It's how disk images, WebDAV, AFP, FTP etc work when you mount them in the Finder. > i plan to write this application in cocoa. Not going to happen. Check out the webdavfs source in Darwin. -- Finlay _______________________________________________ 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: getnameinfo() and NI_NOFQDN
by Marc Majka 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -- Marc Majka On 27 Jul, 2005, at 02:38, Mike Richmond wrote: Mike Richmond Global Graphics Software Ltd _______________________________________________ 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/majka%40apple.com This email sent to majka(a)apple.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… Determining the name of a host is actually pretty tricky, since there are many different names floating around. There are different names that can be set locally. Network name services (like DNS) may have different ideas of the name. If you are mapping addresses to names, different addresses can map to different names. The best API to use is actually gethostname(). It does a lot of smart things, including caching the name once it's found, and it will be consistent with other software that uses gethostname(), so at least you can think of it as a canonical name. If you really can't use gethostname() and want to iterate through all the network interfaces and look for names associated with them, beware that it will cause a flurry of network activity, and that it may take a long time. We've seen a lot of problems with long DNS query times due to old DNS servers simply dropping some queries on the floor. We saw this with AAAA queries in Panther, although it's a lot better now that we run two threads to do A and AAAA queries in parallel. The same thing happens to some people for PTR queries for 192.168.x.x or 10.x.x.x (and possibly other) addresses that are never answered. The result is a long (30 second) DNS timeout. The actual implementations of getaddrinfo() and getnameinfo() are in Libinfo/lookup.subproj, but they just call lookupd (see netinfo/ servers/lookupd/LUServer.m), which does all the heavy lifting. Finally, about NI_NOFQDN: It's a bug in the LUServer.m code. It isn't stripping off the trailing domain names as it should. I am trying to get an unqualified name for the local host. To do this I loop over the addresses returned by getifaddrs(), select the first AF_INET, running, non-loopback address and call getnameinfo() with the NI_NAMEREQD and NI_NOFQDN flag set. I'm expecting the host name returned by getnameinfo() to be unqualified, but I am getting a fully qualified name as if NI_NOFQDN had not been set. I see this on 10.3.9 and 10.4.2. I've looked at the source at <http://www.opensource.apple.com/ darwinsource/10.4.2/bind9-11/bind9/lib/lwres/getnameinfo.c> and that does take notice of the NI_NOFQDN flag. Am I looking at the right code? Am I mis-understanding what the NI_NOFQDN flag does? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Why does free() take 1,100 microseconds?
by John Francini 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Yeah, but... Sounds like a potentially workable solution to me... John Francini On Jul 27, 2005, at 9:54, Markus Hitter wrote: Am 27.07.2005 um 10:41 schrieb Jan Brittenson: Bad advice. For exact this reason. Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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/francini%40mac.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… The earlier lines of discussion made it obvious that this application needs to have near-real-time levels of predictable response. Having the OS decide to page on you when you really don't want it to do so doesn't help, and makes the platform look like it's inhospitable to real-time use. Most ordinary applications don't need that sort of predictable response pattern, but this one does. Allocate it not with malloc, but with vm_allocate, then vm_wire it. This avoids paging [...] If more developers develop such habits, we're back in the Mac OS 9 days quickly. This email sent to francini(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Why does free() take 1,100 microseconds?
by Markus Hitter 27 Jul '05

27 Jul '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Am 27.07.2005 um 10:41 schrieb Jan Brittenson: Bad advice. For exact this reason. Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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… Allocate it not with malloc, but with vm_allocate, then vm_wire it. This avoids paging [...] If more developers develop such habits, we're back in the Mac OS 9 days quickly. This email sent to site_archiver(a)lists.apple.com

1 0