This shift in Python 3 (out in alpha) to a unicode basis, I'm comparing to the move from 16 and 256 colors to millions of colors. The analogy is quite correct, as what it takes to map more colors are more bits, to give a bigger mapping. Same with the space of human languages. I sense a lingering imperialism in some circles that's forlornly clingling to the obsolete notion that all source code of strategic value will stay in the English language or at least in the Roman character set (romanji as the Japanese call it). But I think Python, along with Perl, is doing the right thing in trailblazing a future look and feel. Romanji keywords, OK (that's just part of the syntax in this case), but Lithuanian variable, function and class names, are likewise no problem, at least in 3rd party modules (the vast majority of Python code is 3rd party). We're not there yet, but we're moving in the right direction. I was at one of those Hilton hotel style parties in Bellvue recently (near Microsoft HQS), where some the the guests spoke "beltway" (a curious blend of DC based mumbos jumbo), was chatting with a Voice of America PR person and she's definitely in our camp on this one. Unicode is the way to go. www.voa.gov. Just as English Al Jazeera is taking the time and trouble to reach North Americans in their native vernacular (whether TV execs choose to block distribution is another matter -- Google's YouTube netcasting is a good workaround), so is Voice of America (VOA) netcasting in Urdu. Python's new sophistication around Unicode bodes well for i18n in our gnu math curriculum as well (we make much use of these agiles, as they make the whole field *so* much more accessible -- helps level the playing field). Of course Ubuntu and other Linux distros, Plone, etc. already have a strong track record when it comes to breaking away from "one size fits all" mentality. Kirby