Snappy: March 1st

SUMO Problem Investigation

Looks like increase in negative SUMO reports that I mentioned last time was due to over-counting. There might be some networking regressions, but no general slowdown. Without solid data it’s hard to tell.

In order to provide more context to make SUMO reports more useful to developers we are looking into allowing users to attach their telemetry data: bug 732522.

Work in Progress

There are few exciting landings to report this week, but there is a lot of exciting work going on.

Nick is getting rid of main thread cache locks, bug 723577. Limited initial testing indicates this can significantly improve Firefox responsiveness.

Brian’s prefetch nuking bug 692255 is moving along. Interactivity profiling is also close to landing, bug 710935.

Dietrich is pushing automation changes to how we monitor startup so we can alter our startup sequence to show the main window sooner, bug 715402.

Vlad tweaked chromehang to work on all windows versions and is working on the serverside component with Benjamin, bug 712109.

I posted our plan on shutting down Firefox efficiently, speak up if I missed an important detail. Rafael is working on this in bug 662444.

Jared has landed more of Avi’s smooth scrolling tweaks on the ux-branch, bug 198964. Take the ux nightlies for a spin to try improved scrolling.

Dietrich and Andrew are looking into massive CC pauses after resume, bug 639262.


Async livemarks changes landed, bug 613588. Macro is following up to fix any slowdown for livemark users.

Nathan landed persistent telemetry, bug 707320. It’s not quite working yet, bug 732970.

Local Storage Discussion

Chris wrote a great post on DOM Local Storage, the discussion is great. I did a post on this a few weeks ago.

I have some raw data on exactly how slow DOM Local Storage is, I will post it once Xavier has time to yank data properly out of the cluster.


  1. Great job again guys!
    Love to see the progress being made on these issues.

    I downloaded the UX nightly to try out the smooth scrolling. The long scrolling is nice – where you push and hold and it flies down or up. But the scroll after a single click seems rather slow.
    (Could this be a consequence of an older computer?)
    I’ve noticed that my mom tends to click once & then click again rather than click and hold. So even if she wants to get to the bottom of a page she will just keep clicking. I expect there are other less tech-literate fold who tend to do this as well. It makes the trip down (or up) feel painfully slow.

  2. Thank you for the hard work. any progress on the severe lock ups when flash is used?

  3. I’m still looking for a way to reproduce these on nightlies. We saw this in older versions, but we can’t reproduce it in the nightlies.