[tor-project] PieroV's Monthly Status Report, May 2024

Hi everyone!
Here is my status report for May 2024.

This month, Firefox 128, which will be the base of our 14.0, reached nightly.
Therefore, we're releasing 13.5 soon (with the 115.12 ESR update unless we find some last-minute problems) and then moving all our efforts to the ESR transition.
So, in the past month, I partially worked on finishing some details for the 13.5. The rest of the time, I continued uplifting our patches and rebasing our patchset on Firefox's rapid release to ease the transition.

For the 13.5, I finished the work on the Mullvad Browser Windows installer I started at the end of April [0]. We wanted to make the installation UX as short as possible. Only one click: you confirm you want to install the browser, and that's it :smile:. But if you want, you can still customize the installation options.
When doing so, I discovered the welcome and the finish page of NSIS installers are written with the Modern UI 2 tools you would also use for custom pages. This made the task easier and less hacky, and the final result is identical to what users are familiar with.
I also started implementing a localization pipeline that will be used by Tor Browser's installer as well.

Then, I worked on Android localization. When we evaluated whether we needed to enable multi-lingual GeckoView builds [1], we didn't realize they were also required by the formatting APIs [2]. So, in practice, TBA supported only formatting in en-US. Enabling multi-lingual builds solved this issue.
We also noticed that region data was still leaked. Therefore, I wrote a patch to make the Accepted-Language HTTP header match exactly the tag of the active translation (when not using spoof English) [3].

As for the uplifting work, I finally managed to land a patch [4] I wrote when we were moving from 91 to 102 :tada:. In addition, with the help of Firefox devtools developers, I implemented some tests and uplifted the patch not to persist the data about custom requests done in PBM [5]. This patch was initially developed by cypherpunks1, as well as the patch to add the missing origin attributes to the reader mode [6] that I also uplifted.
The patch for the placeholders of the datetime-local input [7] hasn't landed yet, but at least I found why the tests were failing, and I provided a new revision. I'm waiting for another review from upstream.

For the ESR rebases, I started a branch based on 127 and the branches based on 128, even though it's still nightly.
I also started writing some reviewing guides (which gave me an opportunity for additional self-reviews).
I found a very nice tool called "aha" [8] that converts ANSI escape sequences to HTML. I'm using it to create HTML versions of range-diff, on top of which I'm writing reports on why I needed to adapt our patches.

Finally, I did some of the usual maintenance and help for this month's releases.


[0] Build system installer for Mullvad Browser on Windows (#200) ยท Issues ยท The Tor Project / Applications / Mullvad Browser ยท GitLab
[1] Enable multi-locales also on GeckoView (#41143) ยท Issues ยท The Tor Project / Applications / tor-browser-build ยท GitLab
[2] TBA: formatting APIs are en-US only (#42552) ยท Issues ยท The Tor Project / Applications / Tor Browser ยท GitLab
[3] Restrict the accepted languages to the ones whose localization is available (#42562) ยท Issues ยท The Tor Project / Applications / Tor Browser ยท GitLab
[4] 1787790 - getComputedStyle reports a wrong family for system fonts under certain conditions
[5] 1892052 - The new request data is persisted when done from PBM
[6] 1892046 - The reader mode does not set a first party domain
[7] 1880108 - Placeholders on the datetime widget ignore spoof English
[8] GitHub - theZiz/aha: Ansi HTML Adapter


tor-project mailing list