as soon as you realize you can’t easily contain your commit message within a 50-character conventional message (or slightly more if you wand to be more specific about the scope)
It wasn’t my intention to state that an extensions of certain big software is always better or should get all the credit. No. First of all, I consider Molly protestware and second of all, the thing about being able to do federation and whatnot with much smaller funding was not about Molly. It was about simplex, matrix, XMPP, E2EE for Fedi and handful other decentralized/federated projects. Signal already has been downloaded hundreds of millions of times according to App Store/Play Store and received countless endorsements. And they did in fact face outages after receiving one from Elon Muskrat. So, they needed to find ways to scale better. Their server software could in theory be self hosted, but unlike Matrix or XMPP, it won’t federate so in a way it’s even worse than e-mail when it comes to this. One would thus think that it’s implicit that they would finally add the possibility to let people run their own servers or even devolve towards more P2P-oriented design. But instead they’ve decided to partner with a pump and dump shitcoin scheme whose privacy-friendliness was absolute trash, though granted, that was also at a time when every tech company was trying to join the Web3 hype. Now their reach is even bigger, but has grown at a steadier pace. I won’t try to go more tinfoil here with any unsubstantiated suspicions and begging the question but even though decentralized or federated systems are harder to design in a way that makes them secure, centralized ones are more abusable and create a single point of failure that can affect a large share of the user base.
maybe try setting up a matrix bridge if you feel confident you can secure that properly. On one hand it might increase attack surface (use only servers and bridges with End to Bridge Encryption) but what’s an attack surface on software that is so ridiculously compromised. Also you can try using an alternative client such as Flare. Though YMMV, for me the last time I’ve used it it was quite rough around the edges but I’m happy to see it’s actively maintained so might be worth checking out.
Also no, flatpak doesn’t fix this issue. Yeah it provides some isolation which can be further improved with flatseal, and other defense-in-depth methods. But unless you are willing to face the trade-offs of using Qubes, you won’t compartmentalize your entire system. The key file in question is stored in ~/.local/share. I’m not denying vulnerabilities in userland applications, but thanks to it’s wide reach, often massive codebases and use of unsafe languages like C, it’s the core system or networked software that is the most common attack vector. And that doesn’t ship and will never ship via flatpak.
The most obvious way this is exploitable is directory traversal. But not only that. Just look up “Electron $VULNERABILITY”, be it CSRF, XSS or RCE. Sandbox escape is much easier with this crap than any major browser, since contextIsolation is often intentionally disabled to access nodejs primitives instead of electron’s safer replacements. Btw Signal Desktop is also an electron app.
Also regarding tiling compositors/WMs. Base rate fallacy. Yeah desktop linux has got 3% market share but probably somewhat more if you exclude company or public computers. But then, probably also higher among Signal users. Anyway, that’s probably an Electron issue. Glad to see Flare getting better, so hopefully if it doesn’t get abandoned we might soon have a viable alternative that is more lightweight, secure and integrates better with the system in a more agnostic fashion. Heck, I might be even inclined to contribute a little to that project myself.
yep. Good code is self-documenting and syntax highligting and having longer sections folded up may help more than having to process some greyed out text. But comments are still useful for generating proper autocompletion and avoiding having to skim through you '“self documenting code”. Also it helps greatly with TDD and maintaining good coding practices. For example if you need a numbered list to reliably sum up what some function does, it’s often a good sign that it should be broken into a couple smaller ones.
I see your point and don’t negate such possibility. Although the black box nature of proprietary dependencies in vanilla Signal means an inclusion of potential trojan spyware. Speaking of the need for app lock, as an alternative solution, you can create a separate profile for Signal to have a dedicated PIN. But afaik only GrapheneOS allows notification relaying to main profile. LineageOS on the other hand has a feature called AppLocker. If you intentionally lend your device to kids, Android has a feature called app pinning.
French authorities consider it a “terrorist app”. Louis Rossmann made a video about it. It was in some court case but at this point I don’t remember whether it was a local court or higher and frankly don’t care enough to check.
Have you seen signal’s issue tracker? Ik it’s a big project, but it’s literally getting spammed, plus the desktop app that keeps database key in plaintext and won’t work natively under wayland (needs xwayland, making basic stuff like sending attachments hard if you use most tiling compositor, tho that’s partly Wayland’s design flaw of lacking consistent reference implementation). Also I principally don’t trust apps that rely on both proprietary network services and libraries. The very fact that they don’t leverage their funding to reduce their costs by working on support for federation that is not a matrix bridge (which hasn’t been even developed by them btw) or decentralization, especially since XMPP, SimpleX and Matrix (which has currently 3 well developed server implementations: Synapse, Dendrite and Conduit) have been able to do so with much smaller funding. And it’s Signal, not Molly’s maintainers who have been putting more effort into shiny UX improvements over hardening infrastructure code lately. And even if Signal does improve it’s security, the patches get regularly backported into Molly, whereas even such basic shit implemented solely in Molly, such as app passwords that actually encrypt it’s database is pretty useful. Because even PIN scrambling is not fully immune to shoulder surfing. Defense in deph matters.
tl;dr a longer rant about decentralization vs federation 👇
Even the argument of network effect achieved thanks to reliance on phone numbers is becoming less relevant these days, with DeltaChat providing a convenient way to have encrypted chats using the existing email infrastructure in much more convenient way than traditional PGP. Pixelfed has already achieved E2EE DMs and it’s being worked on for Mastodon. If the UI of the most popular apps and the official web interface are also redesigned to make messaging more convenient to use it might have the same positive effect on user retention as Facebook Messenger once had. Anyway things are bound to change in favor of federation, but not necessarily decentralization. For instance I got mixed feelings about EU’s DMA. I’m optimistic about the interoperability benefits it could bring, but even the official act doesn’t specify how it’ll be implemented. If it relies on something like WebFinger which does require a domain name it’ll end up just grouping a couple of major walled gardens together, so for example SimpleX, Session or Status users still might not be able to chat with people on centralized platforms
You can set up an account over Tor in case of 1984. Haven’t used Orange but mainly due to bigger costs. Iirc the only time my 1984 Wireguard VPN was facing issues was when trying to edit Wikipedia, so not a big problem. Searxng was also working fine.