I just ran across this useful hint in answer to a problem I was experiencing with my Gentoo/PPC installation on my PowerBook G3.

http://www.macosxhints.com/article.php?story=20060814075952448

I recently corresponded with a visitor to my site who was interested in my use of my Cisco Aironet 350 card with my PowerBook G3 Lombard. He was having some difficulty with his 350. In the process of troubleshooting, he discovered that he had an older Ralink card that was supported under OS X.

So, if you have an Ralink card, or are considering getting one, they appear to have fairly decent Mac OS X support.

One point I want to make is that the Cisco card is supported under both OS X and Mac OS 9, whereas the Ralink drivers appear to be OS X only.

Simon, thanks for the info and I hope it helps some other Mac users out there.

I was disappointed to receive an email yesterday from Rob Braun, announcing the soon termination of the OpenDarwin project. As Mr. Braun writes:

OpenDarwin has failed to achieve its goals in 4 years of operation, and
moves further from achieving these goals as time goes on. For this
reason, OpenDarwin will be shutting down.

First of all, I want to thank the OpenDarwin project and those who have contributed their efforts over the lifetime of the project. I have had an opportunity to explore aspects of Darwin / Mac OS X that I would otherwise not have had. I realize that these efforts have largely gone unnoticed and unappreciated. I want them to know that the project will most certainly be missed.

Now, for those that may not know much about OpenDarwin, I’ll try to explain its purpose to the best of my understanding. Apple has been releasing its Darwin source code since Mac OS X 10.0. But this source has not necessarily been very useful. Throughout the development of Darwin / Mac OS X, there have been those that wanted to interact and work with the Darwin source code as they were accustomed to doing with Linux or one of the BSDs. The OpenDarwin project existed primarily to facilitate this effort, to make the Apple-released Darwin code more useful.

So what led the project leadership to the decision to shut down the project? Well, as I’ve written about here before, Apple has recently changed their release schedule regarding the Darwin source. In fact, the x86 sources appear to have dried up entirely. While some have pointed out that Apple hasn’t made any official explanation as to the “delay,” and that we might yet expect the x86 source to materialize, this latest development in the OpenDarwin project casts sharp doubts as to the eventual release of the x86 source, at least in my opinion. Certainly this wasn’t the only issue that was a sore spot between Apple and the OpenDarwin project. Developers have referred to a long-standing distrust between the two parties that stem from various issues. Perhaps this x86 source issue was just the last straw. Indeed the announcement from Mr. Braun indicates a sense of frustration and resignation.

What does it all mean? Well, for many, of course, it won’t mean anything. The OpenDarwin project has lived in relative obscurity for these past years. Certain OD projects, of course, have had great popularity, most notably, the excellent DarwinPorts project. But the core effort to bring order and meaning to the Darwin sources has arguably made little impact in the Mac OS X community. So, while some may not think the demise of the OpenDarwin project significant, I think this may just be another sign that something is going terribly awry at Cupertino.

After all, Apple has benefited measurably from the impression that they are a good citizen in the open source community. I, for one, know a number of colleagues that have “switched” in part because they had gotten that impression themselves, or because I had given it to them. I don’t wish to Apple-bash here, however. Apple is, in fact, an open source citizen. And they have made some important contributions. The newly reorganized WebKit (link will be invalid soon) project shows particular promise. But despite their contributions, I wonder if they can be considered a good open source citizen. And, if they are found undeserving of the credit they’ve received, what then? This post from Mr. Braun describes, in his words, why Darwin failed and serves to shed some light on these questions.

One of the OD project goals, I believe, was to keep Apple honest. If the project leadership feel that they have failed in that goal, what can we now expect from Apple? Is there an opportunity to fill the void left by the OpenDarwin project? Can anyone who enjoys Darwin / Mac OS X hope to participate in or contribute to its development like those who enjoy Linux? Maybe we need a ReOpenedDarwin project.

Tom Yager recently blogged about Apple’s apparent abandonment of the x86 branch of the Darwin kernel source code with the title, “Apple closes down OS X.” (read the article) Ernest Prabhakar quickly answererd with this response, emphasizing that this was really just speculation that Apple would eventually close the source to the (x86) XNU kernel. When I first read this article, I felt that it was a bit sensational myself, as I keep up with the OpenDarwin community, and was aware that it was all just speculation in light of no official word from Apple. In the end, I suspected Mr. Yager was trying to provoke a response from the Mac user community / open source community that might force Apple to finally release the (x86) XNU source.

Well, it seems that his strategy (if this was his strategy) may be showing some fruit. I’m not saying Apple’s showing signs of finally making good on their open source commitment and releasing the withheld source code, but in Mr. Yager’s most recent article on the subject, he writes about Apple’s response to his first article. If you are at all interested in this issue, I encourage you to read this latest article because it helps expose an ugly mindset coming from someone at Apple. What they’re really saying is, “Who cares?” Mr. Yager writes that the people he spoke with at Apple basically challenged him to identify who was really put out by their withholding the source code. They then went on to marginalize the ones that Mr. Yager identified. I think this shows a real lack of appreciation and understanding for open source from someone at Apple.

I understand that Apple is currently facing a challenge in keeping OS X locked to their own hardware, and that the first real threat has reared its head in the form of clones sold overseas. (reference) I don’t know, technically, how to prevent such piracy. Is closing the source an effective way to do it? Perhaps. But I have personally seen Mac OS X (10.4.6) running on non-Apple hardware. If such pirates haven’t had access to the x86 source since 10.4.3, does it seem like they actually need the source to circumvent Apple’s protections? And perhaps there is a way to involve the community in helping to develop a secure anti-piracy / anti-theft component for OS X? Maybe there’s a way to obfuscate the code in such a way as to make reverse engineering the source as difficult as reverse engineering the binaries?

I’m not saying I have all the answers to this complicated issue. I just know that Apple can’t have it both ways. If they want to be a market innovator, and known as a friend to the open source community with all the benefits that come with that friendship, they have to maintain their commitment to open source. I really like what Mr. Yager used to close his latest article:

It’s not about code. It’s about character.

I don’t know what is necessary for Apple to find its corporate character on this issue. Mr. Yager chose to appeal to Steve himself in his article. I will appeal to anyone who identifies themself as being in the “fraction of a fraction” of geeks who are interested in seeing that the XNU kernel source remains open. Post a comment to this blog entry and/or submit feedback using Apple’s general Mac OS X feeback page. Maybe we can all help them find their character by saying, “I do.” when they ask “Who cares?”

I recently read an editorial by Robert Cringely dealing a bit with Mac OS X 10.5 (Leopard), Windows Vista / XP, Virtualization and Windows API compatibility. I wasn’t so interested in his speculation regarding Windows API compatibility as I was his suggestion that Leopard would be sporting a new kernel. In all honesty, I hadn’t ever really considered such a move very likely. I mean, I know people have issues with Mach, but it works, right? In fact, it works quite well. That being said, I’m certainly not tied to it. I began doing a bit of research on this issue to see if I could try to guess what to expect from Apple in Leopard.

The notion that Apple is ditching Mach brings several questions to mind. First of all, why? Well, it appears that Mach does in fact have some significant performance limitations when compared to monolithic kernels. AnandTech put together a nice write-up of its own comparison testing.

So, if you agree that Mach has some significant drawbacks and that Apple is likely to want to replace it, the next question is, “With what will they replace it?” The primary reason for replacing Mach in OS X would likely be for performance, so presumably, the replacement kernel would be monolithic. Of the available open source monolithic kernels, the Linux and FreeBSD kernels present possibilities. I discount Linux for several reasons. First, Linux has some significant technical differences between FreeBSD / Darwin, so marrying the Linux kernel with Darwin could prove technically challenging and unrewarding. Second, Linux has GPL license encumberances. Apple enjoys greater freedom with the BSD license. Of the available monolithic kernels, the FreeBSD kernel seems most likely.

It is entirely possible, however, that Apple has been developing a replacement kernel themselves for some time. I actually think this one is less likely, but still possible. I think it would take an awful lot of resources to develop a separate kernel entirely for all this time.

Another interesting possibility is the L4 kernel. L4 is similar to Mach in its microkernel design, but is meant to address a number of the problems with Mach. There has already been some work done to bring L4 to Darwin and work is ongoing. You can read more about the L4/Darwin (darbat) project here. I’ve downloaded the Qemu image and tried it out, myself. It’s still very early in development, and unless Apple engineers have been working on this apart from the darbat developers, I can’t see Leopard sporting L4.

The last question I want to explore in this post is, “Even if Mach has problems, is it really worth the effort of replacing it?” It seems to me that Mach’s issues have the greatest implication for Mac OS X Server, where things like database and webserver performance are so critical. Well, wouldn’t it be worth replacing Mach to have a competitive enterprise server OS? I’m not so sure Apple is that concerned about it. In fact, here is one account that Apple is not interested in the enterprise at all. So, if Steve isn’t interested in the enterprise market, and Mach is mostly holding back OS X Server performance, does it make sense to swap kernels?

Well, honestly, I don’t know. It seems to me like Mach has seen better days. Maybe the naysayers are right and it is time for a change. Could Apple pull it off? In the middle of the PPC->Intel transition? Yes, I think so. In fact, the time is right for a kernel move. All I know is, Apple’s not saying a whole lot about Leopard yet. This year’s WWDC is sure to bring some exciting news items. Will a new kernel be one of them? I don’t know, but I can’t wait to find out.

Links:
• http://www.pbs.org/cringely/pulpit/pulpit20060420.html
• http://www.pbs.org/cringely/pulpit/pulpit20060413.html
• http://www.anandtech.com/mac/showdoc.aspx?i=2436&p=7
• http://episteme.arstechnica.com/groupee/forums/a/tpc/f/8300945231/m/645003648731
• http://en.wikipedia.org/wiki/L4_kernel
• http://www.ertos.nicta.com.au/software/darbat/
• http://en.wikipedia.org/wiki/Linux_kernel
• http://en.wikipedia.org/wiki/FreeBSD_kernel