September 2009 - Darwin-dev

Re: GCD code in plain old C, and a correctness question
by David Leimbach 11 Sep '09

11 Sep '09

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=gamma; h=domainkey-signature:mime-version:received:in-reply-to:references :date:message-id:subject:from:to:cc:content-type; bh=WiEYRex5b/+b9byNcGJ13o9Ftnn8vutfceW011KW6Ug=; b=rFfYa/qVNxg3Q8ZS+OTnrlQkui5LyXRzjsKa/RMQDgfn9bBl5Nel3K3tEwSC7JnfuO rowmsfSAKAFL/heEm+xY6y7lQFHCgu02J4LDWOI0h4wAKH0rNcCal/w+yTYKWm3MCC1q q+QOhIY/4fKs40e13ND1tX79Et19FyIVM+RpY= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; b=s6jsJateZxBDcnWEDan/aPF7BQ7ZuK1CRcpeQBOalGkdcS6Skg79a90IGmNCRgAjqV RkLAmb/+jyK+Tvy44Zzh4j1bajePKvmWIdLGPsJ5BORR6ZPBp2ce7pKGRT6I9GAfr9Sa AnebarjJu6PAMkfDqA0OpnWJLxJvRX5liMY9Y= _______________________________________________ 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: kqueue and resource forks
by Terry Lambert 11 Sep '09

11 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -- Terry -steve On Sep 10, 2009, at 8:11 PM, Terry Lambert wrote: This is on a normal HFS+ volume, so it's not doing the ._ file You have it backwards. _______________________________________________ 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… One of the engineers here was interested enough in your post earlier to write a test program that looks for events on any file named on the command line. When he ran the program on 10.5, and specified the file name, he got notifications for both file and resource fork modifications. Of course, they were mixed together, and it was impossible to tell which were happening to the file and which were happening to the resource. On 10.6, the same program with the file name specified produced only events for the file. However, if he specified "the file/rsrc", it reported only events for the resource fork and not from the file. So it was possible to differentiate between events on the file and events on the resource, without the values being mixed together for the program receiving the events to (hopefully) separate out. You could go on to consider a file with 250 extended attributes; if you did the 10.5 thing, you'd get an event, then have to check 250 things to see which of those things the event was being reported on; if you did the 10.6 thing, you'd get distinguishable events for the thing on which the event happened. I think this is "as designed", and the resource modification events in 10.5 were a side effect, but we were unable to attribute the change to a particular product requirement (which doesn't necessarily mean anything); you're of course welcome to file a bug on it. On Sep 10, 2009, at 10:42 PM, Steve Gehrman wrote: I just reinstalled 10.5 and XCode and tested it. It definitely worked on 10.5 and now does not work on 10.6 The Finder seems to update, so it must be detecting the change somehow, but I can't get it working on 10.6 using the same code that worked on 10.5 Could you test this over there to see? If you set the file's Open With settings in the finder it will add a resource fork. Set up a test app to watch that file with kqueues and it shouldn't fire on 10.6. I used both my code and Google toolbox's kqueue code. Both were the same. On Sep 10, 2009, at 4:30 PM, Steve Gehrman wrote: I'm pretty sure this worked in 10.5. One 10.6 i set kqueues to watch a file. If I modify the resource fork or add or delete the fork, the kqueue event doesn't fire and I don't get notified to update. Anyone know if this was a change in 10.6? I don't remember getting bug reports for my 10.5 builds. I'm pretty sure this did not work in 10.5, except on UFS, and that doesn't natively support resource forks. In 10.6, we actually added support for a lot of FS-level kevents at the VNOP_ KPI layer, which means that they'll apply to all filesystems equally, but probably still not resource forks individually (part of the removal of UFS support was to support these events for programs that needed them on non-UFS FS types. If you are getting errors now, it's probably because you asked for events before that it was impossible for you to get except on UFS in 10.5, and now you are getting them for HFS+, and aren't reacting correctly to them occurring. Generally, if something doesn't work, you shouldn't be asking for it in old software, in case it starts working in new software. I expect if you were to try this on 10.5 on UFS, you'd see the same sort of things. -- Terry This email sent to site_archiver(a)lists.apple.com

1 0

Re: kqueue and resource forks
by Steve Gehrman 11 Sep '09

11 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -steve On Sep 10, 2009, at 8:11 PM, Terry Lambert wrote: This is on a normal HFS+ volume, so it's not doing the ._ file You have it backwards. _______________________________________________ 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 just reinstalled 10.5 and XCode and tested it. It definitely worked on 10.5 and now does not work on 10.6 The Finder seems to update, so it must be detecting the change somehow, but I can't get it working on 10.6 using the same code that worked on 10.5 Could you test this over there to see? If you set the file's Open With settings in the finder it will add a resource fork. Set up a test app to watch that file with kqueues and it shouldn't fire on 10.6. I used both my code and Google toolbox's kqueue code. Both were the same. On Sep 10, 2009, at 4:30 PM, Steve Gehrman wrote: I'm pretty sure this worked in 10.5. One 10.6 i set kqueues to watch a file. If I modify the resource fork or add or delete the fork, the kqueue event doesn't fire and I don't get notified to update. Anyone know if this was a change in 10.6? I don't remember getting bug reports for my 10.5 builds. I'm pretty sure this did not work in 10.5, except on UFS, and that doesn't natively support resource forks. In 10.6, we actually added support for a lot of FS-level kevents at the VNOP_ KPI layer, which means that they'll apply to all filesystems equally, but probably still not resource forks individually (part of the removal of UFS support was to support these events for programs that needed them on non-UFS FS types. If you are getting errors now, it's probably because you asked for events before that it was impossible for you to get except on UFS in 10.5, and now you are getting them for HFS+, and aren't reacting correctly to them occurring. Generally, if something doesn't work, you shouldn't be asking for it in old software, in case it starts working in new software. I expect if you were to try this on 10.5 on UFS, you'd see the same sort of things. -- Terry This email sent to site_archiver(a)lists.apple.com

1 0

Re: kqueue and resource forks
by Terry Lambert 11 Sep '09

11 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com This is on a normal HFS+ volume, so it's not doing the ._ file You have it backwards. -- Terry _______________________________________________ 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 Sep 10, 2009, at 4:30 PM, Steve Gehrman wrote: I'm pretty sure this worked in 10.5. One 10.6 i set kqueues to watch a file. If I modify the resource fork or add or delete the fork, the kqueue event doesn't fire and I don't get notified to update. Anyone know if this was a change in 10.6? I don't remember getting bug reports for my 10.5 builds. I'm pretty sure this did not work in 10.5, except on UFS, and that doesn't natively support resource forks. In 10.6, we actually added support for a lot of FS-level kevents at the VNOP_ KPI layer, which means that they'll apply to all filesystems equally, but probably still not resource forks individually (part of the removal of UFS support was to support these events for programs that needed them on non-UFS FS types. If you are getting errors now, it's probably because you asked for events before that it was impossible for you to get except on UFS in 10.5, and now you are getting them for HFS+, and aren't reacting correctly to them occurring. Generally, if something doesn't work, you shouldn't be asking for it in old software, in case it starts working in new software. I expect if you were to try this on 10.5 on UFS, you'd see the same sort of things. This email sent to site_archiver(a)lists.apple.com

1 0

New Mac OS Forge Project: libdispatch
by William Siegrist 11 Sep '09

11 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com New Mac OS Forge Project: libdispatch Learn more at http://libdispatch.macosforge.org/ William Siegrist Mac OS Forge _______________________________________________ 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 libdispatch project consists of the user space implementation of the Grand Central Dispatch API as seen in Mac OS X version 10.6 Snow Leopard. The Mac OS X kernel support for GCD may be found in the xnu project. While kernel support provides many performance optimizations on Mac OS X, it is not strictly required for portability to other platforms. However, in order to implement the full API for Grand Central Dispatch, C compiler support for blocks is required. The blocks runtime is available as part of the LLVM project. This project is intended to be a resource for developers interested in learning more about libdispatch on Mac OS X. Contributions to this project will be continually evaluated for possible inclusion in future releases of Mac OS X. The sources are available under the terms of the Apache License, Version 2.0 in the hope that they might serve as a launching point for porting GCD to other platforms. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Darwin Open Source Release
by William Siegrist 11 Sep '09

11 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Aug 28, 2009, at 12:58 PM, William Siegrist wrote: AppleUSBIrDA CF (Lite) autozone clamav xnu http://opensource.apple.com/ Thank you William Siegrist Darwin Team _______________________________________________ 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 Darwin Team is proud to announce the release of the Darwin open source components of Mac OS X 10.6 and Xcode 3.2. They are available at the usual location below. We are still processing some projects, and did not want to hold up the bulk of the release, so the following projects are coming later: The 5 projects mentioned above have been posted, along with the added bonus of CarbonHeaders! This email sent to site_archiver(a)lists.apple.com

1 0

kqueue and resource forks
by Steve Gehrman 10 Sep '09

10 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com This is on a normal HFS+ volume, so it's not doing the ._ file -steve _______________________________________________ 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'm pretty sure this worked in 10.5. One 10.6 i set kqueues to watch a file. If I modify the resource fork or add or delete the fork, the kqueue event doesn't fire and I don't get notified to update. Anyone know if this was a change in 10.6? I don't remember getting bug reports for my 10.5 builds. This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to build va_list on 64bit Intel/darwin?
by Terry Lambert 09 Sep '09

09 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com hi- thanks!- You're looking in the wrong direction by looking for GNU functions. -- Terry _______________________________________________ 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 Sep 9, 2009, at 5:20 AM, lbland <lbland(a)vvi.com> wrote: On 32-bit, I can build a va_list by hand (using simple pointer arithmetic) and call functions that accept va_list. But, on 64-bit the va_list changed and the way I build va_list on 32- bit does not work. Are there some functions that I can uses to push variables onto the va_list (stack)? - maybe even cross-platform GNU functions? ... ANSI-C defines how to pop them off, but I can't find a good way to push them on. -lance About the only use for this would be in-flight hacking of the arguments to a varradic function before passing them off to another, second varradic function, either insertion or deletion. Otherwise you would just make the call to the second function. For almost all printf- like functions you can simple make multiple calls to the underlying function, since they are appending data to an unflushed memory buffer in any case, so there should be no real performance impact since things are being write gathered into the I/O buffer anyway. It's possible to do, but it would be handier to see how you're currently doing it so we could tell you what you were doing wrong. Typically it tends to be a mistake in alignment on the basis of incorrect assumptions about alignment boundaries and packing for a given ABI, typically dealing with type promotion to wider types. It's also why the POSIX standard states that what you want to do is undefined, and why most people go to great lengths to avoid doing it, and why it typically can not be done portably. See also section 15.13 of the C language FAQ. This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to build va_list on 64bit Intel/darwin?
by Jean-Daniel Dupas 09 Sep '09

09 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Le 9 sept. 2009 à 14:20, lbland a écrit : hi- man ffi for details. -- Jean-Daniel _______________________________________________ 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 32-bit, I can build a va_list by hand (using simple pointer arithmetic) and call functions that accept va_list. But, on 64-bit the va_list changed and the way I build va_list on 32- bit does not work. Are there some functions that I can uses to push variables onto the va_list (stack)? - maybe even cross-platform GNU functions? ... ANSI-C defines how to pop them off, but I can't find a good way to push them on. thanks!- If you want to build dynamic function call, use something like ffi which will take care of this for you. This email sent to site_archiver(a)lists.apple.com

1 0

How to build va_list on 64bit Intel/darwin?
by lbland 09 Sep '09

09 Sep '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com hi- thanks!- -lance _______________________________________________ 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 32-bit, I can build a va_list by hand (using simple pointer arithmetic) and call functions that accept va_list. But, on 64-bit the va_list changed and the way I build va_list on 32- bit does not work. Are there some functions that I can uses to push variables onto the va_list (stack)? - maybe even cross-platform GNU functions? ... ANSI-C defines how to pop them off, but I can't find a good way to push them on. This email sent to site_archiver(a)lists.apple.com

1 0