Introduction to the uWolf thread.

This thread is dedicated to discussing the uWolf LibreWolf wrapper, where you can ask questions, help with development, and receive updates on the project.

Due to many bugs LANDSCAPE is officially unsupported.

Useful Links

• The uWolf project is hosted on GitHub and can be found here.
• The latest stable release is available on Open Store.
  

Buy Me A Coffie:

support my ongoing opensource projects, including uWolf here: https://buymeacoffee.com/chromiumos.guy
you have my heartfelt thanks for the donations I received, It means a lot to me and I'm flattered people trust me to this degree.

ANNOUNCEMENT:

as you've seen I have not updated this app in while, this is because I'm working on implementing XDG portals in lomiri this will help all xdg apps integrate smoothly Librewolf is an xdg app as well so this project will benefit.

PLANNED (will come with a massive rewrite i am planning the moment my FP5 arrives and i can develop on platform with VSCodium which will enable me to rapidly debug):

• app confinment (@Fuseteam has an MR that needs to be upstreamed first but I will go ahead)
  there is already a branch which is compliant and contained, heavily broken though.
• fix scaling completely with method @Fuseteam shared
  ui_density=0 equals no scaling issues (haven't tested yet so i can't say for sure).
• rewrite mobile UI (planning on copying fennec's UI close to 1/1 as possible)
  i thought about off-loading UIX changes to an extension but unless i want to do a legacy extension an implementation will take a really long time, so userchrome it is.
• Completely rewrite codebase for Noble/Stability (I now know what sticks so I'll write the code with that in mind)
  everything will get rewritten in chunks, i will conduct some surveys because i have some ideas that are based on some assumptions that are best confirmed before i act out on said ideas.

Development Log (DEVLOG)

Release: v0.11.38

STATUS: Pending Openstore release

• Added codec support: added libavcodec58. (only works in Noble!)

Release: v0.11.37

STATUS: Released on Github

• Librewolf version update: now v143.0.3-1

Release: v0.11.36

STATUS: Released on Github

• Librewolf version update: now v143.0.3-1

Release: v0.11.35

STATUS: Released on Github

• Librewolf version update: now v143.0.1-1

Release: v0.11.34

STATUS: Released on Openstore

• Librewolf version update: now v143.0-1
• Icon Change: chanegd icon to the one agreed on below icon made by AdamSchrey

Release: v0.11.32

STATUS: Released on Openstore

• Librewolf version update: now v142.0-1
• Pathing fix: Pathing is now relative to the $HOME variable instead of a hardcoded username, which will prevent issues if the user's home directory is not /home/username.
• Profile Creation Fallback: Fallback is to create a profile if none are found, negating the setup process of previous versions.

Release: v0.10.31

STATUS: Released on Github

• Librewolf version update: now v141.0.2-1

Release: v0.10.30

STATUS: Released on Openstore

• Generalized path to drm: generalized the path from /sys/class/drm/card0-DSI-1/modes to /sys/class/drm/*/modes as well as made sure the old virtual framebuffer system is used as fallback, which will fix the crashing problem on specific devices.

Release: v0.10.29

STATUS: Released on GitHub

• Fixed is_tablet() Function: The function now utilizes the device-info shell command instead of directly reading YAML files. This enhancement improves reliability and bypasses the fact that AppArmor is blocking access to the YAML files and makes the function break.

• Resolved Profile Change Crash: A significant issue that caused uWolf to crash when writing changes to the profile has been addressed. Although this problem was previously fixed, it required additional confirmation, which is now included in this version to eliminate confusion. The crash was traced back to unusual behavior stemming from the use of both Wayland at the start to obtain screen size with QML and qmlscene before switching to XWayland for the actual application. If this process takes too long, the system loses context and terminates the app. Consequently, if we update the Chrome in the profile, the delay can lead the system to mistakenly believe the app has halted, resulting in a crash.

• Fallback Scaling Method: The old scaling method, which employed Qt grid units and the environment variable $GRID_UNIT_PX, is now implemented as a fallback. This adjustment aims to prevent crashes on devices that do not support the getprop handle for DPI, ensuring a smoother user experience even in scenarios where DPI scaling fails.

Release: v0.9.27

STATUS: Released on Openstore

• URL Bar Fix: The URL bar now hides correctly when entering fullscreen mode.
• Tab Bar Fix: Resolved THE issue where the tab bar was interfering with the OSK overlay, by compensating for it. (the one where the overlay overshoots)
• Keyboard Info Gathering Refactor: The method for gathering keyboard information has been completely refactored. The previous reliance on QML has been replaced with a more efficient Python implementation, streamlining the process.
• Scaling Adjustment: Transitioned from grid units (Qt) to DPI-based scaling, providing a consistent experience across different devices.
• Word Ribbon Support Addition: The OSK overlay now supports the word ribbon feature, accommodating the height of the OSK keyboard with or without it.
• OSK Overlay CSS Refactor: The CSS for the OSK overlay has undergone a complete refactoring, finally ensuring good-ish positioning.
• Usage Mode Fix: Corrected the is_usage_mode_staged() function, which previously defaulted to Staged mode incorrectly.

Release: v0.7.23

STATUS: Released on Openstore

• Librewolf version update: now v141.0-1

Release: v0.7.21

STATUS: Released on Openstore

• Fixed FP4 Scaling Issues: The scaling issue was resolved by capping the maximum scaling at 2.4 and the minimum at 0.7.
• Fixed Tab Bar Issues in Landscape Mode: The tab bar was rewritten to dynamically adjust margins based on screen percentage instead of using fixed values.

Release: v0.7.18

STATUS: Released on GitHub

• Librewolf version update: now v140.0.2-1,
  note: updating to this version seems to fix addon button killing OSK.

Release: v0.7.17

STATUS: Released on Openstore

• OSK Overlay Improvement: Enhanced OSK overlay height calculations to reduce overshoot, resulting in a more accurate and user-friendly experience.

Release: v0.7.16

STATUS: Released on GitHub

• XUI Improvement: Simplified UI changes have been implemented, significantly enhancing the overall XUI experience. Although settings cannot be accessed while in landscape mode, this change mitigates previous issues. Additionally, addon management is now available, and the bookmark button (star) has been reinstated.
• Location Feature: A location policy has been added, providing uWolf with access to location services.

Release: v0.7.15

STATUS: Released on Openstore

• Search Engine Fix: actual solution implemented as suggested by MrOtherGuy here for the Issue here, no more database Injection.
• Librewolf version update: now v139.0.4-1

due to little change in overall architecture no flowchart is provided, for up to date flowcharts visit the github repository.

Release: v0.7.14

STATUS: Released on GitHub

• Simplified OSK Hide Mechanism: This release patches osk_config_trigger.uc.js to improve the responsiveness of hiding the OSK overlay while browsing, making it feel less bloated. The change came from shifting focus to the end-user experience and identifying what triggers the OSK to hide.

Release: v0.7.13

STATUS: Released on GitHub

• Search Engine Fix: Alternative Solution: Although the underlying issue persists, I've implemented a workaround by injecting a pre-populated database containing search engine data. This approach circumvents the need for manual configuration, effectively masking the problem. It's essential to note that this is not a bug on my end, but rather a compatibility issue between LibreWolf and fx-autoconfig. I've opened an issue here for those interested in resolving the root cause.
• Enhanced OSK Hide Mechanism: The On-Screen Keyboard (OSK) hide mechanism has been refined to dynamically respond to URL changes. For instance, when searching on a search engine, the OSK overlay will now automatically hide, eliminating the previous awkward behavior that required manual intervention.

For those interested in the technical details of the database injection and to address the security concerns raised by @domubpkm, I've included a flowchart that illustrates the management of the LibreWolf profile. This should provide a clear understanding of the process and help alleviate any concerns.

Release: v0.7.11

STATUS: Released on github
Significant updates have been made in the release:

• OSK Dynamic Reactions: The browser detection mechanisms have been replaced with a more robust approach, utilizing a Dbus monitor to listen for OSK focus events. However, this new method has a limitation: the Dbus monitor can only detect when the OSK gains focus, but not when it loses focus. This introduced a challenge in detecting unfocus events, which has been temporarily addressed by implementing a crude solution involving a tinted overlay over the screen. This overlay is active when the OSK overlay is shown, allowing the system to detect when to unfocus.
• Bug Fix: The is_tablet function, which resulted in many crashes in the v0.6.10, has been fixed.

For those interested in a more detailed understanding of the new OSK dynamic reactions architecture, a flow chart is provided.

---

If you have any questions or would like to discuss the development of uWolf, please don't hesitate to ask in this thread. Your input and feedback are valuable to the project. Feel free to share your thoughts, suggestions, or concerns, and I'll do my best to address them.

@Fuseteam first let me say this is awesome, second if you want help I want to contribute to this.
I used nixos for while before switching to FreeBSD is has been a joy, so I say this with confidence this is finally the point where we can actually use UT as a full on linux distribution, the influx of packages here will be amazing.

if you don't mind aside from reading up on the code, I will make a UI QML app so end users can just pick and install/uninstall primitive but any UI is better then terminal when walking.

looking forward to seeing a repo on gitlab.

@domubpkm the next step is to wait for Noble, i have somethings i can fix right now
bit i rather just do one big bug sweep session when switching to wayland.

as soon as the transition to noble happens i can automatically make the phone install the subsurface patch (until it goes to upstream) and then i can run wayland which means a host of bugs will be gone but i already know some stuff breaks (and my workarounds for the xwayland bugs) so i will probably need to do a refactor of the entire app so i might as well do everything in one big concentrated push.

edit:
other then that?, really only cleaning up the UI mostly.
i can't really do anything else, though i am toying with the idea of making another startup script to mimic webapp-container so webapps can use this (they'll need to be unconfined)
also finding a way to make uWolf the default browser, after the switch to wayland it should be a lot more usable so it stands to reason people will want to be able to make it open links.

Hi someone by the name of AdamSchrey (on Github) created a bunch of nice Icons for uWolf and I would like to hold a vote on which one to choose from the icons he created the icons can be found here

EDIT: this will be the new logo, hope everyone is happy with the decision

@domubpkm There is a security issue, but it's relatively minor and manageable. The problem lies in JavaScript files with privileged execution that can access browser internals and read files on the host. However, the risk is mitigated by the fact that:

• Librewolf stores sessions in cache, so malicious code would only execute in the next session (after restart)
• The sensitive files are re-read and replaced every time a session is launched, making it difficult for changes to persist
• The time window for tampering with the files is small (between verification of the Chrome folder and Librewolf launch)

To further reduce the risk, encryption and obfuscation of the sensitive files could be implemented, but this would make it harder for me (the developer) to test and debug code. The issue will be addressed once the more significant problems are resolved.

It's worth noting that Librewolf is considered a secure browser, and the privileged code execution is heavily sandboxed, preventing access to web content among other things. While this issue could potentially be exploited to steal data, the overall security of the browser is still robust, and any other security problems would likely originate from the Librewolf/Firefox source code itself.

@Simmonz this is the fix @TotalSonic is referring to (kind of) https://gitlab.com/ubports/development/core/qtmir/-/merge_requests/92
once this is upstreamed firefox and by extension LibreWolf will be able to use wayland directly.

once its upstreamed I'm also planning on releasing my freetube port.

@haagch i'm going on vacation today so development is going to be slower then usual combine that with the NixManager plugin for Noble which im also working on, and the fact i just moved all my devices to Noble in preparation for moving uWolf to Noble proper and i can't really do much about it.

but uWolf is janky like that because of Xwayland if the OSK gives you problems enter the wayland windows (the one with the icon and loading gif) it will redirect you back to the browser but in most cases will also unstick the OSK and make it work again, because somehow Xwayland "unselects" the window that is the browser.

@kugiigi okay to summarize this is the AOSP link to the alerts: https://android.googlesource.com/platform/packages/apps/CellBroadcastReceiver/+/refs/heads/main/res/raw

you can turn off alerts in android with no notice so we don't need to necessarily implement that.

a feature in android which we should also implement is opt-in: "Always alert at full volume" which if turned on will override Silent Mode and still do the alarm.

@AmauryDBZ Hi, developer of uWolf here, on a side note the UI is not modified (in the screenshot) for mobile so this has to be either v0.2.3 which doesn't make any sense or its not loading.
if you could provide me the startup log of the app I would be grateful.

for the spotify problem, disable the extensions first uBlock-Origin and Decentraleyes( which is a tracker blocker).
not to mention LibreWolf by default stops cookies which breaks some sites so you can try to change the security settings and add spotify to the exception rules.

@domubpkm try to reinstall v0.10.30 i patched it it should work now.

I just finished making OpenKav which will be the backbone of an app I'm planning for Israel public transport. and it got me wondering how many Ubuntu Touch users exist in Israel who are also on the official forum, I have seen some mentions in Hebrew forums about Ubuntu Touch so I'm curious.

@lduboeuf the hacky way wpuld be to set it to full then back to original when finished

@lduboeuf no this was consistant through the entire Israel/Iran war, and now.

@sixwheeledbeast yes i meant this when i said standard tone, thank you!

@lduboeuf it should be the standard alert presidential/amber alert sound in my opinion

@kugiigi okay to summarize this is the AOSP link to the alerts: https://android.googlesource.com/platform/packages/apps/CellBroadcastReceiver/+/refs/heads/main/res/raw

you can turn off alerts in android with no notice so we don't need to necessarily implement that.

a feature in android which we should also implement is opt-in: "Always alert at full volume" which if turned on will override Silent Mode and still do the alarm.

I did find a video of the alarm is question: https://youtu.be/sxF2c_nhiGM?si=nXzZYHBnhjPoy-MQ

EDIT:
I think this is the AOSP files for it: https://android.googlesource.com/platform/packages/apps/CellBroadcastReceiver/+/refs/heads/main/res/raw

@kugiigi it is, altough there is a specific alarm sound on both ios and android ill try to look for it and see what the license is, we should preferably use this alarm sound because people will know exactly whats up its very distinct.

EDIT: i haven't actually optted out myself when i still used android but i know it is possible, ill get back to you on if there some sort of standard don't sue us agreement you agree to first because thats sensible to have for once.

im from israel so im intimately familiar with Cell Broadcasts (CB), the problem is that in UT it makes the phone vibrate but gives no alarm so its not useful at night when I'm asleep its not the end of the world but its very annoying waking up to the city wide alarm and having only 5-10m to dress myself and head to the bunker instead if the full 45m for example (the time here is valid for Iran, if it was hamas considering where i live the city wide alarm gives 45s and the CB gives 5-8m). I think adding an opt-out alarm is the best course of action the reason for opt-out is that people expect it and having it off by default may cause them to miss it (very easy when asleep but also music, or being away in another room) and maintaining consistancy with android and ios in this one case is a good idea in my opinion because people will expect it but i'd like other thoughts on the matter.

@MrT10001 yes, but trust me qemu static emulation or box64 can go a long way, not to mention you can just compile it on aarch64. if you use NixManager you can probably get all the compilation dependencies and compile natively on device.