Hi Everyone,
Here's what I worked on over the past week:
I tracked down and wrote a patch for https://trac.torproject.org/18859 Not strictly a TBB bug, but should improve performance for users, I hope.
Also on the subject of performance, I did a small manual test to try to understand the benefits (if any) of enabling HTTP2: https://trac.torproject.org/projects/tor/ticket/14952#comment:24 Indeed it seems the multiplexing feature of HTTP2 can offer us some useful speedups. So at Mozilla I hope to talk to the Mozilla folks about what, if anything, might still be a tracking vector in HTTP2 that we need to fix before we can safely enable it in Tor Browser.
I worked on MPX (https://trac.torproject.org/16352) and made some progress, but it turns out there are more issues with building some of the Firefox components that I need to solve before I have a final patch.
I also spent quite some time on https://trac.torproject.org/23930. I wasn't able to reproduce the problem -- it seems there may be something special in font set that was originally reported by the user. In any case, I have a guess at what is happening, and a possible fix. I'm going to build a Tor Browser for the user to test.
I wrote again to a number of exit operators with bad DNS to ask them to upgrade their tor version. The overall average DNS timeout rate is now down to 4.5%. I'm hoping it will continue to fall and in the near future we can apply a badexit flag to remaining exits showing high DNS timeouts.
So this week I will be meeting with Mozilla and Tor folks for planning the future uplift work, discussing the proxy, browser sandboxing, mobile work, etc.. When I get back I will continue to work on various bits of unfinished business including crashes (#23930 and #24145), #18101, #22343. And I might look into one or two performance-related things.
Arthur