It's interesting because it uses principal component analysis to derive the axes from the data, rather than trying to pick axes in advance. It's also interesting because the most significant axis it finds is the traditional left/right axis, but the second axis isn't the libertarian/authoritarian axis that some assert should be the second axis–rather, it's a pragmatic/idealistic axis.
I've taken the survey twice, once in February and once today. My results were similar: I was in the left/pragmatic quadrant both times. (In February, I was -4.08 on right/left and +5.17 on pragmatic/idealistic. Today I was -3.63 on right/left and +4.65 on pragmatic/idealistic.)
]]>--gc-sections option worked for shared
libraries. (Even working when shared libraries are visible within five
miles would be nice!) The main thing this would help Mozilla with is
getting rid of unneeded copies of constructors and destructors that gcc's C++ compiler emits. And probably
also some unneeded vtables. And maybe a good bit of other stuff.
(gcc could probably be a little better, but it's really a job for the
linker.)DEVICE="acpi" to /etc/cpuspeed.conf), memory
has really seemed to be the limiting factor. (Well, it has a slow disk
too.) I've also been doing some things that use a lot of memory lately.
I wondered why I only got 256 MiB of RAM in the first place—that
didn't seem like something I'd do. So, with the thought of buying more
RAM in mind, I pulled out the folder that had the receipt for my laptop
to see whether I had an empty memory slot or 2 128 MiB DIMMs.
And the receipt said I had ordered 512 MiB of memory. But
free says I have 256 MiB, and the BIOS agrees. And I don't
really ever remember this machine having 512 MiB. So I opened up the
laptop, and, lo and behold, there were 2 256 MiB DIMMs in there. A few
experiments showed that one of the two memory slots was bad. And the
machine is already out of warranty (since I got a short warranty, which
for me is usually a better deal, especially since I'm never sure whether
Dell will honor the warranty after I've installed Linux).
I really need a better memory of what I ordered. Or better memory. Or something. But maybe I'll at least remember not to order laptops from Dell again.
]]>Writing it wasn't nearly as bad as I expected, although I am somewhat
disappointed by the inability of XSLT to re-transform trees (i.e., use
an already transformed tree, in the form of a result
tree fragment bound to a variable, as the source tree for an
additional transformation). I worked around the problem using a named
template and xsl:call-template.
At least I think it's something XSLT can't do—at least I couldn't
see how in the spec.
Most of the quirky behavior of IE for Windows isn't actually needed to display Web sites correctly. It's needed for “Intranet” sites that have been designed to work with only one browser. Backwards-compatibility for corporate, academic, and government “Intranets” is a perfectly good reason to want to produce software that continues to display such sites, but it's no reason to inflict such software on the Web. This would make IE's handling of most web pages much more standard and lessen one of the main ways IE's near-monopoly discourages competition.
]]>