Yesterday WhatWG announced to drop the versioning of HTML. HTML is to be a living standard. This comes at a very inconvenient time I fear.
Over the past few years web standards have, albeit slowly, developed into some
mature state: Most browsers support the current (x)HTML, CSS and JS standards to
the letter and can do more. However, due to the fact that these standards have
evolved so slowly new developments are not included in these standards. HTML5
(W3C and WhatWG)
solves a lot of these problems, including features like
With WhatWG dropping the version number of HTML I foresee a descent back to the Dark Ages where every browser supported a different standard and supported the same standards differently, one browser would support feature X and not Y while the other supports Y but not X. This is a nightmare for developers: not being able to rely on the uniformity of the browsers of your visitors. This created necessities as writing specific CSS and JS code for specific browsers to `fix’ certain behaviour. These fixes mostly had to be done for Internet Explorer as it had at the time by far the largest install base.
Now that we have finally arrived at an era where all browsers support the same set of rather advanced techniques we decide to drop versions. In some cases it is not a disaster to not have a version number. Look at CSS, the changes that have been made to the standard were mostly additions. New operators and new properties and values have been added. These changes required a better parser, and the recognition of the new fields, but never did a semantic change. This, together with the tradition that CSS engines ignore the statements they don’t understand has led to developers being able to include new techniques in CSS files and thus implicitly setting the version of the CSS in the file. As nothing changed older and newer browsers behaved the same at the old statements and only the newer browsers acted on the new statements. So up to now the absence of explicit CSS versioning didn’t cause any problems for CSS.
This changing of semantics without version numbers causes a new nightmare for developers, as they have no way to be sure that their site will look the same across different browsers as they all may support a slightly different incarnation of the standard and on top of that the developer cannot be sure his site will look the same in the same browser a month from now, or even the next day, as the browser can switch to support a newer instance of the standard. As different versions of the same browser may have different engines that support different incarnations of the standard you now also have to worry about the differences between browser versions that your visitors use.
One might notice the presence of a lot of conditions in the above text and think “What is the fuss, it might all not happen.”. The main argument I would like to make here is this: a `living’ standard introduces a lot of uncertainties. At a time where certainties are just starting to come back, giving developers peace of mind, going back to living with uncertainties is certainly a descent into Dark Ages in my opinion.