Internet Technologies Lecture

One student has asked if I could upload Monday’s lecture, rather than just the tutorial. Here’s the manuscript, then. The slides are available in OOo and Powerpoint format.

Colophon: the slides are set in ITC Officina and make extensive use of the open-source Tango icon library. Hands: Gallaudet Regular. Hat: House of Lime. Special appearance as Bunny Rabbit by Sir John Tenniel’s March Hare.

On W3C Validators

I’ve been using the W3C Markup Validation Service since the days of HTML 4.0, the markup specification that came before XHTML 1.0. At some point shortly after, when it became possible to use CSS without causing Netscape’s browser to choke and die on it, I also started using the CSS Validator.

Both services have since become prettier to look at, but I’m not sure the markup validation service has improved a great deal in usability. The CSS validator has always been reliable and straightforward, which is lot more than could ever be said about the markup validator. It used to be a fair bet that wherever the validator pointed out an error, the actual mistake would be a few lines further up in the code from the line that was indicated. The error count used to be grotesquely inflated, too, as any given mistake in the code would spawn an unpredictable number of knock-on validation errors. I haven’t been using the validator a great deal recently, so I can’t really judge its current performance, but the inflated error count still appears to be a problem.

On the up side, though, I did discover recently that the markup validator’s Validate by Direct Input feature will also evaluate Scalable Vector Graphics, which proved handy the other day when I needed to fix a broken piece of vector art.

It’d still be nice if the validator could reliably point to coding glitches instead of requiring all that extra effort interpreting its output.

Some radial graph

Fooling with Graphviz:

Radial graph

digraph G {

graph [root="a",
ratio="auto",
outputorder="edgesfirst",
bgcolor="#eeeeee",
overlap="false"];

node [shape="rect",
style="filled",
fillcolor="white"];

"a" -> "b";
"a" -> "c";
"a" -> "d";
"a" -> "e";
"a" -> "f";
"a" -> "g";
"a" -> "h";
"a" -> "2";
"a" -> "3";
"a" -> "4";
"a" -> "5";
"a" -> "6";
"a" -> "7";
"a" -> "8";
"a" -> "9";
"a" -> "10";
"a" -> "11";
"a" -> "12";
"a" -> "13";
"a" -> "14";
"a" -> "15";
"a" -> "16";
"a" -> "17";
"a" -> "18";
"a" -> "19";
"a" -> "20";
"a" -> "21";
"a" -> "22";
"a" -> "23";
"a" -> "24";
"a" -> "25";
"a" -> "26";
"a" -> "27";
}

Rendered with: $ twopi -Tpng sample116.dot -o sample116.png

This works quite well with just a handful of nodes, but with node numbers in the dozens and hundreds, edge length for the outer nodes increases dramatically. There doesn’t seem to be a way to pack this sort of graph tightly for lots of nodes.

Vector Magic

Judging by the 11,000 bookmarks on Del.cio.us and the 3,000 diggs on Digg, the brand-new bitmap-to-vector service from Stanford got off to a good start. The professionals are impressed, too.

Me? I’ve done some auto-tracing in Inkscape, which does a good enough job for my needs. Still, I’ve just put a number of graphics through Vector Magic that I had sitting around from many years ago: ink-on paper line art of mine that I scanned and reduced to aliased gifs way back in time. Vector Magic did really well on them. Here’s a side by side comparison.

Completely messed up the punchline

View the complete slide show, have fun with open-source icons.