pwnt.be

Adding regexp substitutions like s|<(/?)q>|'&'.($1?'r':'l').'dquo;'|eog just because some browsers don’t like the <q> element is what web design is all about.

I could’ve studied a little more, but creating a multi-color theme for the QDB was just too much fun.

Following a discussion I had over at #bs, I’d like to say a few things about the Semantic Web and the XHTML-CSS combo in particular.
Until recently, I didn’t care about semantics in the HTML I wrote; I would use tables, transparent images, etc. all over and obmit document type declarations. I don’t remember what exactly prompted me to make the switch to writing proper beautiful code, but, either way, it was an accumulation of factors and I had been planning to give it a try for a while.
Having made the switch, I was swiftly confronted with numerous shortcomings of standards—needless to say, Internet Explorer had something do to with some of those. However, after a while, I began to realize that XHTML and CSS are the way to do more with less and that the separation of structure and presentation can in fact be a reality. That prompted me to re-write this very web site. While there’s probably still room for improvement, I think the underlying HTML is now a lot clearer than in the old (table-infested) version and the CSS, elaborate as it may be, reflects the actual rendering rather nicely. More importantly, I can do virtually anything with the layout without touching the HTML or giving up semantics.
Granted, I may have gone overboard. After all, the general user experience on my site has changed little to none. However, I’ve always considered things like the CSS Zen Garden admirable and I’m a firm believer of improving the Web. Even minor adjustments can make a big difference toward user-friendliness or extensibility and especially the ones that take little effort are, in my opinion, always worth taking into consideration.
Finally, if I came off a little strong in the afore-mentioned IRC log, I was by no means trying to offend anyone. I guess—or hope—that’s become my trademarked style of conversation by now.

I just implemented on-the-fly gzip compression here. It was fairly easy, since my (Perl) script uses HTML::Template, so all I had to do—after checking if the client supports gzip, via the Accept-Encoding HTTP header—was find a way to zip the return value of $template->output(). Compress::Zlib::memGzip() gets the job done and it’s about a third faster than invoking gzip. Yay.

Just to give you an idea of how professional local radio stations are … On Saturdays between midnight and 1, on Studio Brussel’s Switch, you ordinarily get Discover the DJ, followed by the DJ Files at 1. Just now, after the midnight news, they played the pre-recorded voice-over for the end of Discover the DJ, immediately followed by the DJ Files intro and its accompanying voice-over and the start of Dave Clarke’s set. After about 5 minutes, they realized they’d fucked up, played a jingle, and Discover the DJ started. In the next five minutes, they managed to accidently play half a station ident right through the music, not once, but twice. Lovely.