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

Re: Experiencing build issues with some libraries
by Dale Walsh 20 May '06

20 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Mta-interface: amavisd-new-2.3.3 (2005-08-22) + Maia Mailguard 1.1.0 at daleenterprise.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On May 18, 2006, at 07:55 , Peter O'Gorman wrote: Yes, I examined the files and it covers both architectures. /usr/X11R6/lib/libfreetype.6.3.dylib: Mach-O fat file with 2 architectures /usr/X11R6/lib/libfreetype.6.3.dylib (for architecture ppc): Mach-O dynamically linked shared library stub ppc /usr/X11R6/lib/libfreetype.6.3.dylib (for architecture i386): Mach- O dynamically linked shared library stub i386 See the "stub" bit, the library contains symbols, but no actual code, like the libraries in the SDKs. See the ranlib manpage. (-c, I think, but I've not got it in front of me). -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.2.2 (Darwin) iD8DBQFEbm3UiD9DTPch4RQRAsQ8AJ42uiqEuoL2vRJPwSo5O1b5ra5e5gCdEZuU 1sTT1Z/C6EnHNJGIBzHv38I= =VIJf -----END PGP SIGNATURE----- _______________________________________________ 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… On Thu, 2006-05-18 at 04:22 -0400, Dale Walsh wrote: /usr/X11R6/lib/libexpat.0.4.dylib: Mach-O fat file with 2 architectures /usr/X11R6/lib/libexpat.0.4.dylib (for architecture ppc): Mach-O dynamically linked shared library stub ppc /usr/X11R6/lib/libexpat.0.4.dylib (for architecture i386): Mach-O dynamically linked shared library stub i386 Thanks for the heads up Peter, I guess it boils down to having a valid X11.root.tar.gz file, it looks like it's been pulled, any way to find someone who grabbed a copy of it before it was pulled so I can get it and continue my building?? Peter This email sent to site_archiver(a)lists.apple.com

1 0

Re: mathlib pow gives bad result
by Lance Westerhoff 19 May '06

19 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Tomas- http://lists.apple.com/mailman/listinfo/scitech It's a good group of folks. Hope that helps! -Lance On May 19, 2006, at 6:56 AM, zahradnicky(a)kppm.cz wrote: Dear Darwin developers, PowerPC-Mac OS X double: 0.44921455671723881 (gdb> print val) Itanium2-HP UX double: 0.44921455671723876 (gdb> print val) Mathematica (PowerPC-double): 0.44921455671723886 649004953142139 (N[val,32]) Mathematica (Pentium4-double): 0.44921455671723875 546774706890574 (N[val,32]) Mathematica (rational evaluation): 0.44921455671723877 615723308414713 (N[val,32]) Itanium2 is obviously closer to the proper result. Should I file a bug against libm? Thanks, Tomas -------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.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/lance%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… I would suggest you also post this question to Apple's Scitech list (see link below). (a) There are actually Apple scitech library developers on the list who can comment/bugfix/etc and (b) there are a ton of people on the list who develop schtech software for any number of architectures and who could comment precision differences. I've discovered that my program works differently on PowerPC running Mac OS X 10.4.6 and Itanium2 running HP- UX. The problem appears to be in the pow function which gives result that differs by 12 ulps from Itanium2. This email sent to lance(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: mathlib pow gives bad result
by Tomas Zahradnicky 19 May '06

19 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com If Itanium2 does x86 floating point arithmetic, keep in mind that x86 operates on 80 bit floats internally, while ppc (afaik) does "only" work with the given 64 bit. So this probably is an architecture issue. You could try and compare this to linux/ppc results (via a ppc/linux live-cd perhaps). If precision is a general problem for you, you might try some software implemented multi-precision floating point numbers, e. g. gmp [1]. By the way GMP 4.2.1 does not compile on itanium2 hp-ux I have :) Tomas -- # Ing. Tomas Zahradnicky, Jr. # The Czech Technical University # Dept of Computer Science, FEE-CTU Prague _______________________________________________ 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 believe that this depends on the mathlib implementation. Some mathlibs such as fdlibm calculate results bit by bit and use integers that they obtain by extracting mantissa from double and extending it to 64 bits thus they obtain internal 80bit precision anyway. This email sent to site_archiver(a)lists.apple.com

1 0

Re: mathlib pow gives bad result
by Markus Weissmann 19 May '06

19 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Fri, 2006-05-19 at 06:56 -0400, zahradnicky(a)kppm.cz wrote: > Dear Darwin developers, > > I've discovered that my program works differently on PowerPC running Mac OS > X 10.4.6 and Itanium2 running HP- > UX. The problem appears to be in the pow function which gives result that > differs by 12 ulps from Itanium2. > > PowerPC-Mac OS X double: 0.44921455671723881 (gdb> print val) > Itanium2-HP UX double: 0.44921455671723876 (gdb> print val) > Mathematica (PowerPC-double): 0.44921455671723886 649004953142139 > (N[val,32]) > Mathematica (Pentium4-double): 0.44921455671723875 546774706890574 > (N[val,32]) > Mathematica (rational evaluation): 0.44921455671723877 615723308414713 > (N[val,32]) > > Itanium2 is obviously closer to the proper result. Should I file a bug > against libm? > If Itanium2 does x86 floating point arithmetic, keep in mind that x86 operates on 80 bit floats internally, while ppc (afaik) does "only" work with the given 64 bit. So this probably is an architecture issue. You could try and compare this to linux/ppc results (via a ppc/linux live-cd perhaps). If precision is a general problem for you, you might try some software implemented multi-precision floating point numbers, e. g. gmp [1]. -Markus [1] http://www.swox.com/gmp/ -- http://www.mweissmann.de/ http://www.opendarwin.org/~mww/ _______________________________________________ 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

mathlib pow gives bad result
by zahradnicky＠kppm.cz 19 May '06

19 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dear Darwin developers, I've discovered that my program works differently on PowerPC running Mac OS X 10.4.6 and Itanium2 running HP- UX. The problem appears to be in the pow function which gives result that differs by 12 ulps from Itanium2. PowerPC-Mac OS X double: 0.44921455671723881 (gdb> print val) Itanium2-HP UX double: 0.44921455671723876 (gdb> print val) Mathematica (PowerPC-double): 0.44921455671723886 649004953142139 (N[val,32]) Mathematica (Pentium4-double): 0.44921455671723875 546774706890574 (N[val,32]) Mathematica (rational evaluation): 0.44921455671723877 615723308414713 (N[val,32]) Itanium2 is obviously closer to the proper result. Should I file a bug against libm? Thanks, Tomas -------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.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

mathlib pow gives bad result
by zahradnicky＠kppm.cz 19 May '06

19 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.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: MacCentral Post
by Shawn Erickson 18 May '06

18 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=beta; d=gmail.com; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=neAxK90fsAbdArpXBeZ/jZGa3MX2OL7WYCzbOJljLdWCcYdp62hhbv/Y2KaYN39UcIpUFW0a8R4G0+VpMkA/7Cmzh1bLrq1BxT+bqzHe3bSHw+QwEMqpVZZJyNTPy0q+sjIBM336UHQYznRMZHyiBzFOGTWuCX4T9jDiooHZ8oo= so we never will see a high performance responding osx? the render speed and so on is really hot on my quad. but os x is responsing very slowly if you look at other systems, like amigaos, morphos or linux. If you want to make any progress (in a discussion on this list or in bug/enhancement reports) on such issues you need to understand, outline, and quantify specifics about your perceptions of performances issues. Mac OS X (like every operating system) is optimized for a set tasks or work flows (ones common to its target market) and so it is hard to do a relevant apples to apples comparison without taking that into account. For example Mac OS X has a generally great latency profile that allows real-time audio processing and playback (not many other operating systems are as good). -Shawn _______________________________________________ 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… On 5/17/06, Marco Thoma <thomasoft(a)ggs.ch> wrote: This email sent to site_archiver(a)lists.apple.com

1 0

Re: MacCentral Post
by Charles Gray 18 May '06

18 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) On 5/17/06, Charles Gray <cgray.darwin(a)cse.unsw.edu.au> wrote: I'm working (with others here at ERTOS) on an experimental port of Darwin kernel to the L4 microkernel. It's all open source, see here: http://www.ertos.nicta.com.au/software/darbat/ Our latest achievements include booting on the EFI release hardware (x86 imac and mac mini), and hopefully we'll have loading of binary drivers working Real Soon Now. We're working on another source release. Thanks! I think so :) Though, at this rate it looks like this L4 kernel has a better chance of becoming the opensource kernel for X86 Mac OS X than the Apple Mach kernel! ;-) Can't hurt! 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… Corey O'Connor wrote: Looks like a great project! Sure, why not? But I hope they make at least a few more code releases for our sake. I think I'll start studying L4... This email sent to site_archiver(a)lists.apple.com

1 0

Re: intercepting system calls works on tiger(10.4.5) but not on panther(10.3.9)
by Peter O'Gorman 18 May '06

18 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Wed, 2006-05-17 at 23:19 -0600, Puthali H.B wrote: > ok, i'm not very much clear on this, so please pardon, if its very > obvious. > > Actually I don't use RTLD_NEXT, but rather get a handle for > libSystem.dylib using dlopen and use that handle in dlsym. And if I use > the resulting function handle from dlsym, shouldn't it call the function > from libSystem, rather than my own? as it does in tiger? Nope, you're forcing flat namespace, so it will still find the first one. I know it's broken, pretty sure that it is documented as broken in the dlopen manpage on panther. > > and will try again with syscall, hopefully it will work. > Darn, looks like a bug in dyld, and one that's not likely to ever be fixed. I see the same behavior as you when I continue on and try to login to the ftp server. So, I booted my 10.2 system and tried there, it works on 10.2, and I'd bet that it works on 10.3.x with x < 4. Long story short, you're out of luck on panther :( Peter _______________________________________________ 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 signature.asc

1 0

Re: Experiencing build issues with some libraries
by Peter O'Gorman 18 May '06

18 May '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Thu, 2006-05-18 at 04:22 -0400, Dale Walsh wrote: > > Yes, I examined the files and it covers both architectures. > > /usr/X11R6/lib/libfreetype.6.3.dylib: Mach-O fat file with 2 > architectures > /usr/X11R6/lib/libfreetype.6.3.dylib (for architecture ppc): Mach-O > dynamically linked shared library stub ppc > /usr/X11R6/lib/libfreetype.6.3.dylib (for architecture i386): Mach- > O dynamically linked shared library stub i386 > > /usr/X11R6/lib/libexpat.0.4.dylib: Mach-O fat file with 2 architectures > /usr/X11R6/lib/libexpat.0.4.dylib (for architecture ppc): Mach-O > dynamically linked shared library stub ppc > /usr/X11R6/lib/libexpat.0.4.dylib (for architecture i386): Mach-O > dynamically linked shared library stub i386 See the "stub" bit, the library contains symbols, but no actual code, like the libraries in the SDKs. See the ranlib manpage. (-c, I think, but I've not got it in front of me). Peter _______________________________________________ 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 signature.asc

1 0