January, 8, 2003 archives

in his answer to the 6th annual edge question, mit professor of psychology steven pinker writes: … little in instructional practice has been evaluated using the standard paraphernalia of social science—control groups, random assignment, data collection, statistics. Instead, classroom practice is set by fads, romantic theories, slick packages, and political crusades. i would note that educational software is much the same, unfortunately. (or at least it was a few years ago, and i assume little has changed since then.) i think an educational software company that actually invested in even minimal research to support the efficacy of its products would find that investment repaid in the marketing value of that information. and they may even end up with better products. (link to pinker's answer via the volokh conspiracy.)

my news aggregation page pops up links into new windows. when those windows are closed (or focus otherwise returns to that original window), safari recolors the links to show they were visited. that's cool.

ah, mark has stumbled into what-if-you-can't-hide-css problem as it relates to safari. in a perfect world, all browsers (and other clients) would implement the css standard (and others) perfectly, and the same. in the real world, they all have little quirks. i consider it a pretty big flaw in css (and similar standards) that there's no sanctioned way to accomodate the real world. instead we have a grotty mess of bugs to exploit, javascript trickery, or server-side sniffing.

of course, then you go down the rabbit-hole of tricking out the user-agent to kind-of claim to be who you're not.

it's great that david hyatt, of the safari team, is addressing this stuff publically. it's too bad there's not just a real public face for the safari team like a shared weblog.

my little aim bot written with Net::AIM stops being able to send messages after some number of hours of operation, and there's no indication why. the bot is still connected, still gets (and responds to) keep-alive requests from the server, can receive messages, and doesn't get error messages when sending messages, but the messages don't go through. very annoying. i fixed the Net::AIM code to roll over sequence ids properly, but that doesn't appear to be the problem (and i don't think that event has ever even been reached).

the yahoo and icq bots don't appear to have this problem.

i'm not sure which would be more useful—a method to tell a particular browser that it is the only one who can use a rule, or a method to tell a particular browser that it should ignore a rule. i think you need both. with the former, you can apply additional rules to work around a bug triggered by earlier rules. with the latter, you can prevent a rule from triggering a bug.