As far as I can tell IE9+ should support the features needed to layout the word cloud. Unfortunately though it does not work out of the gates.
I have spent some time trying to figure out why, but haven't managed to get it working properly. The problem seems to be related to the getImageData method on the canvas. My tests revealed that in IE9 it was consistently returning all zero values even after the fillText method had been called.
Add DOCTYPE for IE9.
This is required to ensure IE9 runs in "standards mode".
You should make sure IE9 is running in standards mode, e.g. by setting a DOCTYPE. I’ve fixed the example so it does this, and I’ve tested this in IE9.
This doesn't seem to fix the problem for me. Have you tried the Twitter example at www.jasondavies.com/wordcloud/ in IE9? The docset declaration is there and IE9 defaults to standards mode but only a single word appears.
Yes, and it worked okay for me. What is the exact version of IE9 that you’re testing, and OS? I saw references to a bug limiting the number of pixels for getImageData to 8192 in IE9, so perhaps this is affecting it.
Turns out that it only doesn't work when running Windows 7 with Parallels. When booting directly into Windows it works fine. Weird.
Perhaps it’s an outdated version of IE9 with the getImageData bug?
It is a single installation of Windows 7 + IE9 that works fine natively but is broken when virtualised. Possibly some strange anomaly in Win API when virtualised. I am just going to ignore it.