What Does X-UA-Compatible Mean For Me?

So here I am a couple weeks after the IE team announced through a variety of different channels their proposal to help cushion the blow of their next browser release through the use of a META declaration and HTTP header, “X-UA-Compatible” describing what browser(s) the page and its associated styles and javascript files target.

I’ve got plenty of thoughts on why vendor extensions and related adjustments to behavior are bad. I also have some concerns over this one in particular. Extensions, and more general workarounds and hacks of all kinds [vendor driven or not] get buried into code, reused, copy and pasted, dropped in application templates, removed by accident, and generally used by people who don’t know why they’re doing it, but instead just because it works. As anything other then a very temporary, one time use solution this doesn’t seem to me to solve any problems that are inherent with either web standards or an ecosystem where content publishers are open to who they let see their content. But let me not get too far off onto that tangent and consider first that the proposed solution goes through.

For the moment I want to focus on the practical — what does the additional rendering mode and the ability to switch to it via META declaration mean to me as a working web developer?

There’s been lots and lots of discussion of the proposals via blogs, mailing lists and other forum, but they’ve focused on either how the proposal was raise and whether its appropriate direction for the technologies web developers work with to take. Through all those discussions however I’m still having trouble wrapping my head around the long term impact of the proposals and how I need to manage my very non-theoretical code and projects. In a search for discussion of these practical examples on how one would start to tailor their code in this new model I’m turning to you all. Can the extensions, and the additional rendering mode they’re implying be worked around with no impact at all, or if it requires new planning on the part of developers, or is it just more of the same planning we’ve been doing for ages.

Right Now

In an XHTML1.0 IE6 & IE7 world my typical structuring of code to ease browser hacking and targeting has been as follows:

  • use an XHTML1.0 [transitional or strict] doctype that triggers “not-quirks-mode”
  • start code with no browser hacks
  • For combined IE6&7 adjustments [hasLayout type stuff usually] use a separate style sheet linked inside conditional comments
  • For IE6 but not IE7 use the * html hack when needing to hide stuff, or remember a few extra rules that anyone can get safely [display:inline; float:left;]
  • Ignore IE5.x and less
  • Keep more complex CSS and other hacks to a bare minimum [e.g. clearfixing, etc.] and instead design and code as ‘universally’ as possible

That has worked for the last 1-2 years in a world that knew nothing about IE8.

With IE8 in our sights

But now we’re in a world that knows IE8 exists which changes things just a little bit if I want to be a responsible web developer. Any document I publish or help a client publish from today on I have to consider [with limited knowledge of details] if and how I want to prepare them for the upcoming browser changes.

  • Do I leave my above solution in place and do nothing? In doing so I have to hope that IE8 will be good enough to not get IE7 hacks but act like Gecko or WebKit — that or reside myself to expecting that some adjustments will have to be made once IE8 lands.
  • Do I prematurely start to include the like-IE7 meta tag so that I don’t have to revisit the site later and just let the site live forever as IE7 compatible? Is it realistic to expect that this will still work, and there will not be any testing when IE8 hits required, and that things will work flawlessly without going through the same motions that should happen with any new browser release?
  • Does it make a difference if I think the site will have regular staffing and updates or if its just a site build to get something published and nobody manages it again until the next redesign in a year or two?

Reasonable approaches all — I may flip flop some while I decide what I think works best and see what others are doing. However, they’re also solutions for the extreme near term. Looking further out the answers and options start getting fuzzier.

After IE8 hits

Shortly after IE8 hits web developers will have to switch from asking themselves how to code for the unknown, isntead having to code for the new known. It is easy to anticipate that a requirement for any new site in this time will be to cover exsiting compatibility [IE6 & IE7] and extend to the new kid on the block. So how might we do that with all these rendering modes floating around? At the pace standards move along, as well as the need to avoid total screwy quirks mode in the older browsers lets presume we’ll need to stick to the same languages that we’re using currently [for some time at least]. So to revisit my current model of structuring code for compatibility’s sake out 2 or 3 years what would that look like?

  • Do we continue to play in IE7-standards-mode, coding down for IE6 and hope IE8 needs no adjustments and get nothing out of the IE8 progress?
  • Do we use no compatibility flag and presume IE7 compatibility mode and then code up or down as needed and cross our fingers?
  • Do we use the use the IE8 compatibility flag and code down to both IE6 and IE7?

What about JavaScript?

One thing we, as web developers, haven’t had to deal much with — ok, I lie, we’ve had to deal a lot with — but anyway, IE8 promises to have some large javascript feature changes and it has been cited [or presumed] that this as a big reason about why the separate “mode” is needed. Though I welcome the progress on the scripting front, the “rendering modes” is not an angle that has yet impacted JavaScript code. What real impact if any will the same browser and version changing javascript behavior based on mode have on our existing or future sites? Will it have any impact at all? Is the last option for browser version management feasible if the JS changes are indeed drastic?

We may not know how to answer this until some concrete information on JavaScript changes in IE8 is posted, or even still not until we see a public beta build released to the public. That said, I’m not sure the question, nor the answers are premature if we’re being asked now to evaluate and embrace the proposal.

Thoughts?

In a roundabout way I’m trying to both air, but also think through, my concern that the middle term outlook [2-3 years, or as long as XHTML 1.0, IE6 and IE7 all live] the additional rendering mode of IE8 and mode switching mechanics will keep developers in a “think like its IE7” mindset and not take real advantage of IE8 because of the difficulty of mixing the IE8 benefits with something that would work reasonably well in IE7.

I think its great to have a handle on the short term solutions, and its certainly enjoyable to envision a future with HTML5, CSS3, SVG, Canvas, DOM3, XML, XSL, and whatever other standards may be in the works. But should I be concerned that we’re setting ourselves up for a /worse/ middle term outlook then we would be if we kept on as we have been? Am I way off base with these concerns? Have I totally missed the point or the timeline?

Are there potential solutions or coding practices I haven’t thought of?

Read, Reflect, Opine — IE8 Rendering Switch Proposal

Through ALA and the IEBlog comes a proposal for a new mechanic of handling the backwards compatibility of web sites — pushing the familiar DOCTYPE Switch aside and going with a new mechanism of declaring target browser versions via meta tag.

The perfect solution or the last sign of the web standards apocalypse? If you’ve got an opinion let it be heard.