Monday, 20th January 2003
Stylesheet parsing gets complicated
Craig Saila points to the SearchEngineWatch Webpage Size Checker. It’s a nice tool, but it doesn’t appear to take the size of linked style sheets in to account. I was playing around with the idea of a web page cache written in Python over Christmas and I hit the same kind of problem—while finding linked stylesheets using Python’s HTML parser wasn’t too difficult (and could be achieved equally well using a regular expression) things get a lot hairier when you start to take @import statements and CSS defined background images / custom bullet images in to account. Again I imagine a solution could be hacked out with regular expressions but a nicer method would be some kind of CSS parser (the Python standard library has yet to include one). Maybe another project for a rainy day...
Scaling the two way web
Another Dave inspired post: It seems I misunderstood Dave’s objections to blogging feedback mechanisms yesterday. I thought he was ruling out what I see as an invaluable tool for low traffic bloggers, but in fact his main complaint was that things like comments / TrackBacks and so forth simply don’t scale. Mark Pilgrim echos his complaints, pointing out how overwhelmed his blog was with auto-linkbacks from his Safari review. l.m.orchard concurrs: At present, I’m safe. My rating is Mostly Harmless, so all my open systems are mostly free from abuse. But, the first time I really strike a nerve somewhere, I’m a sitting duck.
You know me
Dave Winer: The “You Know Me” Button. Dave hates posting comments on blogs and then having to check back constantly to see if anyone has replied (I do too). Sam Ruby’s solution is to provide the comments as a separate RSS feed for each of his entries, but Dave wants something more automatic that won’t clog up his aggregator. Dave’s new proposal is intruiging to say the least. When you sign up for an account with a discussion forum you have the option of configuring a link to an “identity server” able to respond to a specific protocol. Once this has been done, the discussion software “pings” your identity server with your username and a message whenever someone responds to one of your posts.
[... 402 words]