evolt has an article, DHTML Text Marker - An Experiment, with an interesting concept: do that highlighting like Google does on the client side. That would reduce the necessity of, eg, the little app the Perl Bookshelf CD has to do text searches and highlights in the results. As proprietor of my own personal bookshelf CD, that'd be nice, but I haven't thought of that in a while at all--I'm not even sure what platform I'd write the app to.
Meanwhile, it links to this classic summary of the argument for and against the innerHTML property. I would say this is an important consideration for me, building that Mozilla WYSIWYG widget thing (shhh, remember?), but since Q42's code already (and rationally) builds everything from nodes, I'm only using it to read the HTML off. If I didn't use innerHTML for that, I'd have to walk the node tree, building my own HTML string. That'd be OK, but since Q42's code already uses Mozilla-specific CSS properties, I may as well not bother being more standard.
