I still use an Ipod Gen5 with RockBox, because a) it works and b) I get to use an open source firmware, which means I don't have to worry about whether $BIG_VENDOR has bothered to support OGG/FLAC/etc files.

Admittedly technology is moving on, but from the standpoint of a device that does one thing and does it well the older Ipods with RockBox do just fine. Why upgrade just for the heck of it?

Heck, i've still got an old iRiver T30 tucked away somewhere that takes AA batteries, which I'm not inclined to get rid of either... small, functional, and does the job.

As computer technology matures, hopefully we'll start to see at least some boutique shops crop up whose goal is to make the IBM Model M keyboard equivalent of things like music players - I'd gladly pay a premium for a device engineered to last 30 years instead of 3.

Weird as it sounds with all the electronic label printers you can get today, there's just something about the old style "punch the label as a 3D letter into tape" approach that I prefer. Especially when the tape punch is a serious tool, not those cheap plastic versions:

http://makezine.com/review/too...

Perfection in engineering... it not only solves the problem of creating the perfect typing experience, it's also tough enough to use as your own personal Hammer of Thor when your office mates storm your cubicle trying to stop the noise.

Let me guess - are you a Cubs fan?

IRC is still used as a major form of (semi) real time collaborative tool in free software development. Freenode remains hard to beat for this purpose, and I don't really see it changing anytime soon. It's not so much a question of not giving it up as seeing no compelling reason to replace a (very nicely) working solution to the problem.

I am intrigued by these statements:

"Within Google we use a custom build system with a custom build file format."

"...at Google there is no distinction between "builds on my machine" and "builds on someone else's machine."

I'd be curious (if you can say much about it in a public forum) how your build system compares to the approachs used by tools such as CMake and Autotools in terms of introspecting into systems to determine what headers, functions, etc. are available. We use CMake and try to detect all of the characteristics we need to know about the host operating system environment at configure time. How does Google's system tackle the problem of cross operating system portability? Say, for example, someone needs to add Haiku to the operating system targets for their project? What is the process?

I still remember the fascination from when I first watched The KGB, the Computer, and Me. It was many years later that I finally read The Cuckoos Egg, and I found that even more enjoyable - a fascinating story, well told. I still have it on my bookshelves today.

I also have one of the Klein bottles - a very nifty product, entertaining and educational at the same time.

Thank you for making such rich contributions to the world.

Thank you for that article - awesome and interesting reading.

I'd really, REALLY like to get my hands on a powerful Linux laptop with such a high resolution screen... if I could afford it I might even settle for the virtual machine solution on the Mac, but a full-up Linux laptop with such a screen would be ideal.

During certain kinds of software development, it isn't uncommon to accumulate a dozen or more terminals and application windows displaying relevant content. Given good eyesight, there simply is no substitute for a high PPI screen when doing such work. Ditto for studying high resolution photos or working with computer aided design. If I could find an affordable IBM T221 monitor with the right adapters for modern graphics hardware, it would STILL be superior to anything I could buy at consumer PC monitor retail. (Unfortunately, the adapters and setup are apparently a tricky proposition even if you can find the monitor.)

I've looked now and then, but I still haven't been able to find any indication of when PCs will begin offering high PPI displays, or even whether the rest of the computer industry is *trying* to catch up with Apple in this respect. Has anyone seen any hints?

If we have to wait for "proper" OS support, they'll never come - the OS support won't be fully fixed until there is a demand for it. And the higher cost/lower yield for high PPI display production means it's a risky, difficult task to try boostraping the market from the manufacturing side.

I'm hoping a hybrid approach might be workable - at SIGGRAPH a few years ago, Microsoft was presenting work on technology for splitting a display signal up over multiple screens. If a way could be found to mount multiple iPhone-type screens into a monitor configuration and translate input over them, that might offer a viable way forward.

High density PPI displays are extremely expensive to produce because of the zero-defect-over-large-surface-area manufacturing issues. Since iPhone screens are smaller and already being produced in large numbers, it might be more practical to splice a bunch of those together. Edge visibility when "stacked" is probably the greatest physical hurdle - I'd guess it's a toss up between that and the inability of current graphics cards to drive such a monitor for "biggest practical hurdle."

Still, if one manufacturing process could turn out vast numbers of small screens that can either be used for phones or assembled into monitors... that seems to me like the only viable approach to the "too expensive to manufacture" problem you face with things like the IBM T221.

Acceptance of textbooks - verification of quality and being usable in courses - is a big hurdle. It's the equalivent to the problems open access journals are currently striving to overcome - building a reputation takes time, and without it viability in education/academia is a difficult proposition. Hopefully this is a useful way to start building momentum - I think it would be an excellent way to get more educational value for the dollar.

If the idea can gain broader acceptance, there are a number of interesting ideas for open source textbooks that can be tried. I like the idea of developing a K-12 and college educational plans in an "open source" fashion, identifying the resources needed, and mapping out the missing pieces as a guide for where to concentrate efforts to create open source teaching materials. I suppose there is only so much you can do to "solve" the problem of what constitutes good teaching materials, since that will vary with learning style, cultural and linguistic background, etc. but it would be nice to have a systematic framework and forum within which to try, evaluate, and evolve ideas. It will be interesting to see whether, as open source devs age and start having families, interest in open source educational materials also grows - that's when the question becomes directly relevant and worthy of resource commitment for a lot of people.

Ideally, open source materials could be managed for "dependency satisfaction" - i.e. mastery of the material in grade 6 materials associated with a project provides the necessary and sufficient foundation to learn grade 7 materials, and so on. That was sometimes a frustration for me growing up - structured resources with fine-grained pointers like "understanding of this concept requires understanding of X, Y and Z" from previous years would have been nice. Sort of "knowledge building a.l.a math proof."

Even for public domain recordings?

If they're complaining about having to compete with a more advantageous cost structure established by the non-profit for royalty requiring songs, I take it there would be no objection to the CBC streaming public domain and Creative Commons licensed content? (I'm assuming Canadian law doesn't mandate royalties be paid for any playing of any content, but that's an assumption - somebody please correct me if I'm wrong.)

On a broader scale, I sometimes wonder if we need to have a public conversation about the fundamental motivation for allowing and promoting non-commercial activity, and what kind of society we really want to be. As I understand it, non-profits get treated differently because they are (theoretically) providing some benefit to society at lower cost than would be necessary to an organization performing the same service while trying to make a profit at the same time. Lower cost means greater public benefit for the same resources committed, and that greater public benefit is valued more highly by society than the specific lost opportunity for someone to make a profit.

In principle, if you disapprove of non-profit activities, couldn't it be argued that the very existence of ANY non-profit is unfair competition to some potential for-profit company? Do people who think this way see any value in anything that isn't tied to profits? Are municipalities that want to provide public internet to all at low cost as a utility (information becomes just like power and water, not an unreasonable analogy given the way our society currently functions) doing something wrong? Are libraries ruining the commercial market for books and other consumer media? Are museums wasteful institutions because they lock up artifacts that could otherwise be immensely profitable as commodities being bought and sold in the art and collectibles markets? Is public schooling a bad idea because it competes with private schools that would otherwise be able to pick up the business? Where and how do we draw this line?

it's how you handle it that counts. Years ago, I was part of a program where a college did some summer school programs for (IIRC) middle school students designed to give them more exposure to science. On the whole it was a good program, but the college physics students working that summer looked at the physics questions on the final test and discovered several problems. To the credit of those running the program, when the college students pointed out the issues to the program leaders they either struck the questions or gave credit for correct answers when more than one answer was shown to be correct. And they did so as the test was in progress, rather than let the students trip on them and get slowed down. I was impressed at the time, and am more impressed in retrospect.

Science questions can be tricky to get right - what seems like an unambiguous question when it is written turns out to be much less so when you start thinking more "generally" about things like frame of reference. It's important to own up if those kinds of mistakes happen though, because the students who are thinking about the questions deeply enough to spot those issues are exactly the ones you most want to encourage in scientific study. The response "yeah you're technically right but we're not changing your score because we meant this" is very discouraging, and will tend to cause students to shy away from complex subjects. It demonstrates that learning the material is not always enough to get decent grades - why bother putting effort into it when there are other fields that more reliably reward their efforts?

Part of me wonders why teachers are still having to write their own questions for basic subjects like this... you'd think there would be Creative Commons licensed materials assembled that had been widely vetted and community reviewed... add a bunch of vetted, correct "twists" to each question that the teacher could opt for when assembling a given test and memorizing all the possible answers gets prohibitive - or at least, gets hard to do without actually learning what needs to be learned to answer correctly in the first place...

I'm using this driver (well, probably a slightly older version of it) with my desktop now, and so far I've been pleasently surprised. I don't need blazing fast performance on 3D for most things. FlightGear/OpenArena level games are about as far as I'm likely to push, since I'm not into the latest and greatest FPS anymore. Given that, the prospect of an integrated driver that "just works" without having to do anything extra is awesome.

My last Gentoo re-install I ended up trying the Nouveau driver after my attempt at enabling the binary NVIDIA driver didn't go well - had to flip on a couple kernel options to get acceleration, but after doing so and for my uses the results are "fast enough." I'll be sticking with Nouveau from now on unless I hit a major show-stopper. Well done, Nouveau team!