Quoderat

[Update: Tim O'Reilly writes that not only Apple but Red Hat should be worried about Ubuntu.]

We bought our older daughter a cheap used ThinkPad as a grade 8 graduation present, and I installed Ubuntu Linux on it for her. It was by far the easiest install I’ve ever done for any operating system — everything just worked, including sound, wireless networking, DVD playback (once I downloaded special restricted packages), remote printing, and suspend-to-ram. I did not have to edit a single text config file — I could do all of the setup, even WiFi, entirely through the Ubuntu/Gnome GUIs. [Correction: I did edit /etc/apt/sources.list to add the sources for the restricted packages, but I think I could also have done that through the GUI if I'd wanted to.]

Why not a Mac?

My daughter is familiar with Linux (at least the GUI parts), MacOS X, and Windows XP. We had considered buying her a Mac notebook, but used Macs cost more than new Wintel books of the same capabilities, so that was a non-starter. She doesn’t much like Windows, probably because they make her use it at school, so Ubuntu it was.

I always liked the Mac interface, especially during the early period up to about 1990 when they led the industry in GUI innovation. Contrary to the accepted wisdom, however, before MacOS X I never found Macs to be a particularly stable desktop computing platform, even compared to Windows. For example, earlier versions of MacOS had highly unreliable TCP/IP support and required huge fixed-size memory allocations for applications, problems that other OS’s had long ago fixed — as much as I liked looking at Macs, I always shrank in terror when family members asked me to help fix problems with them. Windows might crash a lot, but at least it crashed in more predictable ways.

Return of the king …

Fortunately, Apple finally addressed these problems a few years ago by admitting that their backend was completely broken, throwing it out, and replacing it with Unix in MacOS X. They’ve even admitted that maybe the one-button mouse wasn’t such a great idea after all. After that the Mac, if still overpriced, was a strong and stable platform: Mac notebooks started to reappear at IT conferences, Mac ports of Open Source software flourished, and everyone in the Mac world was generally happy. Mac even started innovating in GUI design again, after letting Windows take the lead for a decade.

… but the natives are restless

So why are some of the highest-profile Mac users starting to show disaffection? Mark Pilgrim was the first to announce that he was moving to Ubuntu Linux, and he recently posted a very funny list of Ubuntu essentials for ex-Mac users. Next, Tim Bray announced that he was thinking of switching to Ubuntu, though he’s worried about WiFi support and LCD projector support (my experience has been the opposite of Tim’s). Now Cory Doctorow is also planning to move from Mac to Ubuntu. For all of them, I think, one of the big issues is gaining control over their data which is now locked into Mac proprietary formats that Apple changes at whim.

Cool or cheap?

Do only three defections matter, even if they’re high-profile? Possibly not for Mac usage, but perhaps for Mac price. Historically, Apple has been able to sell its computers for up to a 100% price premium because of the perception that they’re cool — if the mavens suddenly decide that Ubuntu is cooler than MacOS, as seems to be happening, Apple’s price premium could suddenly evaporate.

The end of the Mac-is-cool myth would be as good for Mac users as it would for everyone else (except Apple itself), since cheaper Macs could mean a much larger user base. Would even the hardest-core Mac afficionado complain about a $500 Mac notebook? I didn’t think so.

Today is the submission deadline for XML 2006 papers and tutorials. They’ve been pouring in over the weekend, and I’m very happy with the high quality of them. If you have not had a chance to submit yet, please send in your proposal today.

We’re getting close to the XML 2006 paper/tutorial deadline and there are a lot of good submissions. Please keep ‘em coming:

http://www.idealliance.org/xmlusa/06/call/

If you’re proposing a paper, all you need to provide right now is an author, title, and abstract — the full paper (in PDF or XHTML format) won’t be due until 18 December, after the conference (Boston, 5-7 December).

For more details, see the call for participation. And thanks in advance to everyone who volunteered to referee papers — you’ll be hearing from me soon.

Next Monday (19 June) is the main paper and tutorial submission deadline for XML 2006 (Boston, 7-9 December; Tutorials 6 December). You can submit proposals for papers and tutorials through IDEAlliance’s system here:

http://www.idealliance.org/xmlusa/06/call/

For more information, see the full call for participation.

Tim Bray updated an old piece on binary search this morning — I missed it the first time around, so I was glad that it popped up in my blog reader. Tim’s taking some flak about data abstraction from people who don’t have his experience in high-performance environments, but what got my attention most was his mention of using gaps in a long array to provide efficient updates.

It turns out that this technique, called a gap buffer [wikipedia], is one of the cornerstones of text editors like Gnu Emacs. I’ve been using Emacs for 20 years and have contributed to the main distribution (see derived.el), but never bothered to look at the C code long enough to discover this particular technique. There’s surprisingly little information online — if anyone’s ever bothered to do testing for the optimum gap size, etc., it’s not showing up in Google — but it’s still nice to experience the joy and excitement of a new (to me), simple algorithm that solves a common problem well.

Does anyone have pointers to more detailed research on gap buffers? It seems to me that they’d have applications far beyond text editing, including (perhaps) storing compiled tree data (aka binary xml) on disk.