HTML 5 Thoughts

I was reading through the HTML 5 draft (actually glancing … since it goes on and on). If you're looking for a summary, I think the HTML 5 Differences draft is more useful. As I read through the list and the TOC, I had a few different reactions.

There are definitely some bits of goodness in it, albeit a few years delayed, and those additions will realistically only become available and dependable still more years down the road. Its great to see the <input> tag finally grow up to support common semantic data types such as date, time, url, email etc. It is also good to see HTML incorporate media scenarios by adding the <video> and <audio> elements as part of the tag soup. Other useful additions that I think will go a long way are better and standard support for application scenarios such as managing navigation history, local storage, drag/drop, drawing and content editing. There is definitely lots of possibility, esp. when combined with the latest CSS capabilities.

The spec is long! And I wonder if it should even be called the HTML 5 spec. It looks like it wants to be an end-all spec for the browser platform, and even speaks to socket connections and peer-to-peer connections. However, at the same time it doesn't speak to other foundational UI and presentation features such as animation, or data-binding. Instead it adds a whole bunch of new tags like <dialog> which to me seem less critical at this point in time (I'll blog about this some more ... see here).

The spec doesn't seem speak to extensibility either. To me this is a key feature, and personally near-and-dear, having worked on the platforms side of software for so long. Back in May of '06, I listed my wish list for the browser platform... and one of the bigger items on my list was getting DOM fundamentals right. Specifically, I would have loved to see the HTML 5 spec address extensibility of the tag set and rationalize things like HTC behaviors and XBL bindings. We now have built in media tags, but what if I want a <slideshow> tag? It simply doesn't address encapsulation of behavior, composition, layout extensibility, and ability to define new object models.

Its over-encompassing nature, especially beyond HTML in a strict sense, also makes me wonder if there shouldn't be a much more trimmed down HTML 5 Core spec. I'd love to see something that has a chance of being realistically and fully implemented in the foreseeable future.

In the meantime, it looks like browser plugins such as Silverlight will continue to be an important part of the RIA platform, by bringing not only consistency in APIs and capabilities across browsers, but also by nailing some of the fundamentals.

What are your thoughts? Or your rants? :-)