Create a gist now

Instantly share code, notes, and snippets.

What would you like to do?
Things to do when IE9 Beta comes out

Things to do when IE9 Beta comes out

Note significant features added since IE9 PP4

  • Implemented but not announced: box-shadow, hsla
  • Expecting: 2d transforms, 3d transforms, css animation, css transitions, flexible box model
  • Maybe: gradients, columns, reflections, svg filters, uncrippling @font-face
  • Maybe version auto-update functionality

Test Modernizr against it

  • Does it still throw an error checking elem.msTransform?
  • Are test results consistent with advertised support? ticket
    • Areas of intrigue: flexbox, transforms, indexeddb
  • Any substantial things worth adding into Modernizr 1.6 as a result of IE9 support?

Update CSS Hacks article

  • Word on the street is there won't be a way to target IE8 but not IE9.
    • That would suck. Someone find one!

Update CSS3Please

  • Add any new supported values, prefixed or unprefixed.
  • Will they kill filters?
  • We want to get rid of the IE8 filter syntax as IE8 reads the 6,7 variant just fine.
    • Unless IE9 doesnt have transforms but does have filters and only reads the -ms-filter variant.

HTML5Boilerplate stuff?

  • Hmm. Everything should be peachy.
  • Maybe double-check conditional comments setup.
  • Verify IE=Edge forces documentMode == 9

HTML5Rocks updates

  • Maybe add new support for prefixed transforms in the studio?

Explain @font-face caveats with IE9

  • The situation is: if you do not have a WOFF in your declaration.. and are using the bulletproof (smiley) syntax, you might have a problem.

  • Again, If you have a woff in there, you're fine.

  • The TTF has this little fucker called an embedding bit which has four possible values. IE9 requires the TTF to have it set to 0, which means Installable. Chances are your TTF file doesn't have it set to that.

  • If it's not 0, IE9 will not subsequently grab the .eot that <= IE8 typically grabbed.

  • You can solve this by either now serving a WOFF.. or by adding an EOT to the primary src descriptor:

     @font-face {
       font-family: 'Graublau Web';
       src: url('GraublauWeb.eot');
       src: local('☺'),
              url('GraublauWeb.eot') format('embedded-opentype'), 
              url('GraublauWeb.ttf') format('truetype');

Plan for deferred celebration

Added by @dshaw:

Celebrate that 3 years from now we'll be able to share a modern web experience with the IE user base. #heyyo

Word on the street is there won’t be a way to target IE8 but not IE9. That would suck. Someone find one!

God I hope someone comes up with a solution… If IE9 doesn’t fix those CSS parser bugs, this will break existing sites relying on IE8 browser hacks.

As a last resort, I filed this as a bug on Microsoft Connect. Please vote there if you would like this to be fixed.

Aww you orphaned ! It needs some love too!


paulirish commented Sep 12, 2010

Added, divya. <3

If you're using paul irish's body conditionals, it doesn't matter if IE9 can be selected with css hacks. It's invalid anyway so stop it. =)


paulirish commented Sep 12, 2010

Well then I'll document that. :) Good suggestion.

@timmywil: Of course, but still it would be nice if existing sites relying on IE8 browser hacks wouldn’t break in IE9, right?

Keep up the good work Paul. We all really appreciate it. :-)

asrail commented Sep 13, 2010

@mathiasbynens: actually, are you saying that CSS hacks should be versioned? They already have conditional comments.
Hacks are fun, but they're just hacks. Nothing more than it.
This way people learn they should make the things the right way.
Why can't someone fix a bug regarding selector or comments? No one should rely on third-party bugs to build their programs.

People will file bugs (or mail the site owner) and they should use conditionals. They'll have to change anyway, so make it the right way.
The only valid use of IE css parser bugs that I've seen was to differ IE 5 on Mac from IE 5 on Win. All other should use conditionals.

@asrail Not that discussion again…

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment