I’m looking for an android browser to open webapps isolated from my general browsing
I’ve been using Mull as disposable browsing and brave for login into places (shopping, social), but I don’t like that if I want to stay logged it also keeps history record. I think it would happen with any browser I use, right? Any suggestions?
There are 2 answers to this.
Security
From a security point, Firefox on Android does not isolate processes and not even itself in a good way (I will try and search up the exact reason, its some modern Android feature that Firefox supposedly does not use).
So we should assume that Firefox is less secure on mobile.
You should not use it, but use Cromite or Brave, Vanadium on GrapheneOS. Note that most FOSS “privacy browsers” that have a download size of under 100MB will use the system webview and also not be able to isolate processes.
Usability/Privacy
I use Mull and Librewolf on my Devices.
Use Mull, disable keeping history (but do not enable “delete cookies”) and use “Cookie Autodelete” to replicate what Firefox desktop can. You set it to delete all cookies, but you visit sites where you want to stay logged in, open the popup and whitelist only them. You “outsource” the cookie cleaning to the addon, as FF mobile doesnt have this feature.
I do it like that on mobile and desktop, delete all cookies and only keep those where you want to stay logged in.
On Mull also install UBlock origin. If you want security and an opt-in approach, install NoScript too and set “default” to not allow any javascript. You will need to “unbreak” every site you visit.
This approach will spare you of hundreds of embedded javascripts on websites, and you manually allow only what seems okay. (You mostly never know if it really is, as Javascript is often obfuscated). This is good for privacy and security.
the theory behind this
Adblockers and malware scanners use “badness enumeration” which means “allow everything but block a, b and c”. This is fundamentally flawed, as malware can easily change “how it looks” (encode and decode again, or use randomized obfuscation) and with ads you will always have to keep track of changes.
The list of malware and ad sites will grow and grow, slowing down machines and consuming tons of processes.
Noscript and the cookie approach are the opposite, you block everything and the list you keep is only as big as the stuff you want.
Please also read the more detailed words on some points, but note that I am not an Expert
I am not sure but Firefox desktop (gecko) and mobile (fenix) are separate projects. I think they share a lot of code though.
So a browser renders webpages, CSS, Javascript and nowadays even Webassembly and WebGL stuff.
If you stick to HTML and CSS you are way more secure, Javascript is way more attack surface, and JIT+Webassembly where about 50% of Microsoft Edges security vulnerabilities.
So if you disable WebGL and JIT+Webassembly by default the attack surface is way lower. But there are websites, especially good stuff like Tuta doing in-browser encryption over advanced quantum resistant protocols, that require JIT. GrapheneOS Vanadium now has a toggle to enable JIT for some sites, Edge Desktop has too afaik (dont use it but its likely the most secure browser on Windows), Chrome-ium desktop allows a policy where you can list JIT exceptions.
Firefox desktop and mobile both likely allow blocking wasm and jit in about:config but no GUI toggle yet. Torbrowser in “very secure mode” also blocks JIT.
Coming to the engine, Chrome-ium focuses a lot on sandboxing, while using C and C++ for like everything.
Firefox has the rendering engine completely rewritten in Rust.
Assuming that most sandbox escapes come from memory issues, firefoxes sandbox doesnt have to be as secure if they simply dont have memory issues as they use Rust.
Firefox Desktop uses Seccomp filters for every process (restricting syscalls the processes can make) and separates processes using unprivileged user namespaces for every process (I think they call that fission). Firefox Flatpak can only use seccomp filters but the processes are not isolated from another, which is why you should probably use a system install (binary, .deb, .rpm, …).
You can sandbox system installed apps using bubblejail and allow access to the syscalls needed to create user namespaces. But it is pretty complicated and incomplete in my experience, even though I dont know why.
On Android firefox still uses the engine in Rust afaik, but Android has a very different model how to isolate apps. Also only Chrome-ium can isolate every process strongly on Android, yet. It is totally possible but Mozills doesnt seem to care.
At the same time to be honest I never had a security vulnerability in Firefox affect me, my entire life.
It is important to protect activists and people that “StAtE ActOrS” want to target, so those people should use the most secure browser possible.
The problem is simply that we have no privacy respecting variant on the desktop, that also has quick CI/CD updates. Ungoogled Chromium often lacks behind on Updates (and dont mention Thorium please).
This could be done by people that know how to build such a pipeline on Gitlab, Github etc. and apply all the ungoogle patches to the browser, hardening the build and creating .RPM packages every few days. There just is nobody currently doing that.
Removed by mod
What do you mean by this comment?
Removed by mod
I am not parrotting their Propaganda, see my other, longer comment. I attached it to the top one now, and changed the catchy phrasing.
See this GrapheneOS Discuss thread where they deleted or have hidden my comment.
But as you seem to not have read the probably more detailed other comment, please do so.
I dont know what you want to tell me with the “mail thread about hating tom” as I have no idea what I just read.
I guess the beginning of the conversation and especially Tom Ritters mail here most useful.
Removed by mod
I dont get that forum. They dont actually recommend Brave, only some “lower level” people do. Some people that parrot what the others say are often mods and take care of blocking everything. A post I made about that issue was also hidden, I can still see it and likely the mods.
Especially bad if in the same thread misinformation posts are kept. But deleting them is also not fair, as they are partly right.
I understand that you have no time to write about that in detail, but they do. Apart from the Torbrowser thread I read nowhere about the key differences in sandboxing and site isolation.
I also read nothing yet about how “leaky” Chromium is in terms of security. For sure Torbrowser doesnt accept it as it permanently contacts Google. But this is about security. I dont use Chromium anymore after finding out how much it pings Google using googerteller.
But just saying they have superficial opinions like that is not fair.
If you have sources that prove me wrong please add them. I asked Firefox Devs about the security of Flatpak Firefox, also because it seems they dont do anything against their “Firefox is insecure” reputation. For sure, big corps like Microsoft and Google are pushing Chromium, but they simply also add a lot of informative sources for research.
Bullshit. Please cite where this should be the case .
Tbh today was the first day reading a mailing list and its very strange. Their tracker, and the duplicate of that are both nearly empty and have no info, the only info is from that mailing list.
So it seems you are a mod, and if you think I need to do that I want to have at least another mod look at this.
I am critical of GrapheneOS, as clearly demonstrated, but your comment didnt disprove anything I said but you just ranted about how bad they are. I dont think this is very constructive.
Removed by mod
Thanks for the link. I will read through the Tor mailing list and the other parts.
Why would you say something like that? But I will check everything again.
This was about Android, and Firefox had no process isolation to my last knowledge. Looking to find a source against this currently.
Okay? This is about privacy and security, Chromium is fully open source. Yes it is big tech and that may be a problem in many many cases. But this has no effect on its security.
You mean using brave is better than firefox on Android?
No, not as a general statement. Read the comment again
Maybe you should elucidate why you think this is the case, Rather than replying dismissively.
Frankly, I don’t trust what you say because you only make a claim - you don’t back it up with any explanation.
Check your hubris.
I too am curious about the real world implications of this lack of process isolation.
I am not sure but Firefox desktop (gecko) and mobile (fenix) are separate projects. I think they share a lot of code though.
So a browser renders webpages, CSS, Javascript and nowadays even Webassembly and WebGL stuff.
If you stick to HTML and CSS you are way more secure, Javascript is way more attack surface, and JIT+Webassembly where about 50% of Microsoft Edges security vulnerabilities.
So if you disable WebGL and JIT+Webassembly by default the attack surface is way lower. But there are websites, especially good stuff like Tuta doing in-browser encryption over advanced quantum resistant protocols, that require JIT. GrapheneOS Vanadium now has a toggle to enable JIT for some sites, Edge Desktop has too afaik (dont use it but its likely the most secure browser on Windows), Chrome-ium desktop allows a policy where you can list JIT exceptions.
Firefox desktop and mobile both likely allow blocking wasm and jit in about:config but no GUI toggle yet. Torbrowser in “very secure mode” also blocks JIT.
Coming to the engine, Chrome-ium focuses a lot on sandboxing, while using C and C++ for like everything.
Firefox has the rendering engine completely rewritten in Rust.
Assuming that most sandbox escapes come from memory issues, firefoxes sandbox doesnt have to be as secure if they simply dont have memory issues as they use Rust.
Firefox Desktop uses Seccomp filters for every process (restricting syscalls the processes can make) and separates processes using unprivileged user namespaces for every process (I think they call that fission). Firefox Flatpak can only use seccomp filters but the processes are not isolated from another, which is why you should probably use a system install (binary, .deb, .rpm, …).
You can sandbox system installed apps using bubblejail and allow access to the syscalls needed to create user namespaces. But it is pretty complicated and incomplete in my experience, even though I dont know why.
On Android firefox still uses the engine in Rust afaik, but Android has a very different model how to isolate apps. Also only Chrome-ium can isolate every process strongly on Android, yet. It is totally possible but Mozills doesnt seem to care.
At the same time to be honest I never had a security vulnerability in Firefox affect me, my entire life.
It is important to protect activists and people that “StAtE ActOrS” want to target, so those people should use the most secure browser possible.
The problem is simply that we have no privacy respecting variant on the desktop, that also has quick CI/CD updates. Ungoogled Chromium often lacks behind on Updates (and dont mention Thorium please).
This could be done by people that know how to build such a pipeline on Gitlab, Github etc. and apply all the ungoogle patches to the browser, hardening the build and creating .RPM packages every few days. There just is nobody currently doing that.
Pinging the people in this thread (another advantage of commenting and not just liking). Plase see the changes in the above comments.
You should definitely not use Brave under any circumstances. If you care about security, then using a spyware (which is what Brave is) is not a good option.
https://spyware.neocities.org/articles/brave