Google's crazy ID and class naming conventions

I was just playing around with Firebug and editing Gmail's CSS file, and I wanted to edit a button, but the div ID for that button was :rj. I am fairly certain that CSS does not allow colons in —and especially starting as—ID and class names. So my guess is this is some advanced trickery. I'm not sure if it's consistent like this for each user, but FWIW, the ID was for the "Search Mail" button at the top of the page.

Does anyone know what they are doing an开发者_开发问答d how they are doing it?

IDs used to be quite strict on what was allowed, not so much for classes. HTML5, however, has lessened a lot of the restrictions on what could be ID values.

Here's an article on the what's allowed for IDs and classes in HTML5: http://mathiasbynens.be/notes/html5-id-class

It's enough to make your head hurt.

EDIT:

To address more of your question about why Google is using a seemingly random ID, I'm sure the IDs and classes they are using make perfect sense to their programmers.

Google only owns one browser so if they want to create a web app that is cross browser compatible like firefox they must follow the same standards and rules as every other web developer. That said most modern browsers are pretty lenient when it comes to parsing html so while your example is not correct HTML and probably wouldn't validate there would be no visible negative effects.