Other Firefox-related browsers also had segfaults a couple months ago. Disabling the LTO build option resolved the issue.
This can be closed after a user reports the latest version of the package works. If no one confirms in about a week, I'll run a test build.
I would use wlheadless-run (aur/xwayland-run-git). The problem is how much effort it's worth to defend if someone stubborn complains. Across a few browser packages, I recall seeing only two...
Some comments about the revised AUR package:
- Should declare variables that users are expected to be able to change with following format so that they can be set in environment. (Sets...
This is resolved for Chaotic AUR by setting with-app-basename explicitly to Firefox to make clear it is intentional.
There are other Firefox forks that use de-identified user-agent regardless...
After turning off resist fingerprinting (RFP) in official .deb downloaded from OBS, linked from...
This PR is still misguided, especially with how emphatic the comment is: "DO NOT USE!" Would be more acceptable if it said "Avoid use." However, if resist fingerprinting (RFP) is turned off,...
If ResistFingerprint is turned off, that's technically not a "break", but intended behavior.
User agent appear unstable, but not clearly related to --with-app-basename. See discussion at #2235
# https://arkenfox.github.io/TZP/tzp.html
Mozilla/5.0 (X11; Linux x86_64; rv:135.0) Gecko/20100101 Firefox/135.0
# https://whatmyuseragent.com/
Mozilla/5.0 (Windows NT 10.0; Win64; x64;...
This PR is misguided. I just installed chaotic-aur/librewolf 135.0-1 built on Fri 14 Feb 2025 07:41:12 AM UTC and a few other forks that set --with-app-basename. The user-agent was not...
I just installed chaotic-aur/librewolf 135.0-1 built on Fri 14 Feb 2025 07:41:12 AM UTC. Default profile. Checked user agent at https://www.whatismybrowser.com/detect/what-is-my-user-agent/
``...
Downsides would depend on alternatives. The practical choices are to use the known workaround or disable PGO. In principle, if extensions can be kept active during profiling, their performance...
Reason for performance difference between packages is most likely from PGO. Last year, I benchmarked many browsers compiled with different settings using speedometer 2.1. PGO had a 15-20%...
FYI, 115.14.0 recently released. If you're having difficulty building, I found the following needed to be downgraded: Rust 1.77, LLVM 17. To prevent errors during PGO, some patches are needed.
The .hg folders aren't used and don't affect the build. I suppose it doesn't matter much if you have the spare bandwidth.
Looking more carefully in the tarballs, seems a lot of space can be saved by removing the l10n/*/.hg folder. Uncompressed, they take up 943/3019 MiB. They are not used during build.
Also,...