For mapping optimization tiers of frames over time.

We discussed this being a stacked mountain chart[0]. The most important thing to convey for JIT optimizations over time is hotness over time (% of samples in a range that the selected frame is executed out of total samples in a range), and what tier, if any, the frame settled in. While stacked mountain graphs do not easily show which group is greater, and can be visually ambiguous, this in itself conveys that there is not one obvious tier the sample was found in. Additionally, with the "hotness" conveying total samples observed that the frame was the leaf, aggregating the count regardless of tier, this shows which areas it matters that the tier settles into ion. The data source expected for this is [{ x: time, ys: [percentage, percentage, percentage] }, ...], with each percentage (0-100? 0-1?) representing the peak of a group for that bucket. Bucket/windowing conversion from sample data to plottable data is handled in bug 1174889. [0] http://2.bp.blogspot.com/_D9-JNTtRKgs/TC9RiVb3tMI/AAAAAAAABEE/ES1tv8H8R6k/s1600/Picture+165.png

Comment on attachment 8624407 [details] [diff] [review] v1 Review of attachment 8624407 [details] [diff] [review]: ----------------------------------------------------------------- Looks good -- should make a followup bug for a bit more comprehensive test and `setTheme`