Some people have asked what HTML mode I finally settled on. I'm using a customized version of Daniel Pfeiffer's adaptation of James Clark's sgml-mode. Yeah, there's lots of SGML stuff that I don't use, but it has a set of HTML keybindings and is easy to add things to. For example, here's my code for inserting an image from Amazon:
(define-skeleton amazon-image "HTML image tag for an Amazon book" "ASIN: " '(setq input "") "<a href=\\"http://www.amazon.com/exec/obidos/ASIN/" str "/windleyofente-20\\"> <img src=\\"http://images.amazon.com/images/P/" str ".01.MZZZZZZZ.jpg\\" border=\\"0\\" align=\\"right\\" hspace=\\"5\\" vspace=\\"5\\" /></a>" )
You can do much more complicated things. For example, I like to build tables of image thumbnails for use in blog articles from time to time as in this example from WWW2006. I created a picture-table function that codes up these tables with relatively little fuss and they're just the way I like them.
I think it's fine to talk about some product being a great editor and having good support for language Blub, but I want more than that. My editor has to be customizable to exactly fit my needs. I can't imagine a non-programmable editor being flexible enough to provide significant support for that task. If your editor isn't programmable, you're missing out!