Monday, 3 June 2002

More on the browser feature I want. I'm tempted to learn how to hack a Mozilla sidebar and write this, but I barely have enough time to breathe right now, and I still haven't got that sodding Java RCS up to dogfood yet. Number of visits shouldn't be the only deciding factor on what goes on the list. A site I've been to several hundred times in the past, but not at all in the last two months is probably less important than one I've only been to three times, but all in the last week. Perhaps a rules-based scoring system of some kind. If a page doesn't make it on the list in the first pass, the feature should start aggregating pages and combining their scores, either by making use of <link> tags to identify which pages are related (and which one is the "first" page that should end up in the list), or in the absence of link tags, guessing parent URLs. Dmoz.org should publish an XML-RPC API, so that URLs in the list could be looked up to see if they're in a particular category. It would be better to have this feature, and have it be wrong one time out of three, than to not have the feature at all. 11:51:25 PM

I hate keeping bookmarks. Often, I'll find myself having visited a site a lot through links from other sites, but never remember its URL because each individual time I visited, the site wasn't important enough to remember. I want the web browser to keep track of what sites I visit frequently, and put them in a list for me. And I want the feature to be smart enough to work without me having to perform any configuration, or maintain any lists manually. The bookmark paradigm hasn't really changed substantially since Mosaic was the Cool New Thing. Considering how central the concept of "remembering where we've been" is to the whole web experience, you'd think the bookmark/history tools would have evolved a little more than they have. 3:06:11 PM

© Copyright 2002 Charles Miller. Last update: 19/6/02; 4:04:27 PM.