HTML5: Spec Is Great But Browsers Make it Pointless

Ok, pointless is obviously too strong a word but here's my beef. I'm familiar with HTML5, the new elements, API, etc but I wanted a deeper understanding so I started reading HTML5 in Action. What is becoming a common theme is "...to support browsers that don't yet support all of HTML5..." which generally means writing a decent amount of JavaScript. So if I'm writing the JavaScript to validate forms for browsers that don't support the HTML5 built-ins then what's the point? Just use a jQuery validation plugin and move on.

I realize HTML5 is still a moving target and not finalized (was 4 ever really finalized?) but I'm finding it difficult to really grab hold of it at this point, beyond a few things, because it isn't really saving me any headaches. That could be another one of my misconceptions; HTML5 isn't meant to make my life easier. I'm looking for some points/guidance to validate strong implementation of HTML5 in existing web applications. Thought this might be a good discussion.

P.S - I'd really like to avoid discussing the failures of HTML as a whole, Mark.

I use HTML5 boilerplate and don't worry too much about older browser support (IE6 users? Bug off!)

The modern browsers have enough support for the HTML5 elements that I use that I don't find myself frustrated too often.

Bear Bibeault wrote:I use HTML5 boilerplate and don't worry too much about older browser support (IE6 users? Bug off!)

The modern browsers have enough support for the HTML5 elements that I use that I don't find myself frustrated too often.

Yea, I use that as well. But it isn't just IE6 we're talking about. Most of the good stuff is only supported in IE10. 8 and 9 are still heavily used. Also, FF still doesn't support some of the input types. "Modern browser" is a fantasy term meaning "what we use to develop with" In the real world you know there is no such thing as a modern browser.

I was hoping for more conversation than "use this library". formnovalidate and formaction. I don't see anything in Boilerplate that deals with this and you need to write a small amount of JavaScript to deal with it. Yea, that's a small one and probably not heavily used. But they add up.

Hmm, perhaps you are doing more advanced things than I am. I haven't really run into many problems with what I've been trying (mostly new semantic elements and geolocation).

Well if you work on an intranet or for users forced to use certain browsers, you don't have this problem

. Of course, then you have absolutely nobody on IE 10 so it becomes a moot point

.

Gregg: How is this situation different than when CSS evolved though?

We are always going to be stuck with older browsers that do not support x and y. We are used to the server where when we upgrade to the latest we have everything, sucks that the client is not like that.

Where I work now, we built our own library that adds a lot HTML5 support for things we use. The framework handles the detection and the developers just write the markup like they should. I am willing to bet some of them have no clue that browser X does not support Y. We are slowly adding support for all the HTML5 stuff and hopefully some day we can open source it.

The answer is if you want all browsers to behave the same way, it will take a lot of code. Sad thing is not all the browsers do the same thing for HTML5 controls and it is basically impossible to detect those differences without user agent sniffing and pior knowledge.

Eric

Eric Pascarello wrote:The answer is if you want all browsers to behave the same way, it will take a lot of code. Sad thing is not all the browsers do the same thing for HTML5 controls and it is basically impossible to detect those differences without user agent sniffing and pior knowledge.

Eric

Exactly, so why even worry about those types of "features" in the spec?

Jeanne Boyarsky wrote:Gregg: How is this situation different than when CSS evolved though?

I think it is different because it is more than just styles. There are new elements, new attributes, new JavaScript API's, not to mention CSS3. Beyond that, it probably isn't much different. I just wasn't doing a lot of it when that happened so I don't really know for sure.