• 1 vote
    GitHub provides a service called "GitHub Security Advisories". Some projects i use which aren't even developed on GitHub, use this to publish security notices (and do coordinated disclosure across ecosystems, i think?). This perhaps makes sense as an authoritative place to publish details about security problems, and several other release channels would link to it? i'm vaguely under the impression that this might also show up in people's feeds if they have the GitHub repo starred, and it'll be probably be attached to automated scans of container dependencies (at least, so long as they point to ghcr)
  • 1 vote
    duplicate of #3454? i like the idea of being able to anonymize messages instead of deleting them. but i don't like the idea that server moderators should be able to force you to only anonymize, never delete. in particular, why are you leaving that server and deleting your messages? probably because you don't appreciate the community anymore. i feel like you should probably be allowed to withdraw the value you have provided, at least on a platform level. see also: Fandom, which doesn't let you delete a wiki even if the entire team of contributors moves to an independent place, and Reddit, which has forcefully replaced the moderation team of subreddits that have protested website admin decisions. in both of these cases, the people in power are kind of holding your contributions hostage, as if they own those contributions. i don't think Fluxer should be providing those kinds of tools to communities. they are already regularly abused.
    This poses an issue to servers that are a support group for specific bots, games and other stuff
    in the specific case of support groups, there is an endless pain on the internet of seeing forums be replaced by discord servers. because now you have to log in just to read the messages, and they're not nicely structured or indexed in a search engine. Chat apps like Fluxer generally aren't a good place to host documentation and if that's your use case, please use something else. it's for chat. while permanence is nice where all parties consent to it, chat is inherently ephemeral. see also: media attachments on Fluxer having an expiry date. Fluxer's forums may change this? those are vaguely supposed to be publicly webindexable? maybe in that case, it can make sense to have forced-anonymization? but idk what those will look like yet. take note of the fact that Fluxer Feedback is a distinct website and not just a forum channel in Fluxer Labs. this matters for the usability of it as a support forum. way more, in my opinion, than the chance that some people might want to withdraw their contributions.
  • 1 vote
    CNR on new beta. Version 2026.11.100, Beta 6, Channel Beta, Build 3020, Android 17 (arm64), Device Google Pixel 9a, Locale en-US, Push unifiedpush, Flutter 3.47.9, Dart 3.13.7 it's fixed
  • 1 vote
    fixed in the latest beta. notably, clipboard history (which my video doesn't showcase) also works fine now; i can paste whatever even if it's not "actually in my clipboard", and from the keyboard quick bar, and it's parsed properly. i can even paste, like, "half a mention" and it properly enrichens. (e.g. <#15 + 58585104790462467> hello properly becomes a channel mention, even though my cursor was never at the boundary of >) it feels like it is now impossible to get the message box into a state where it contains unparsed markdown. at least from testing with mention entities specifically. however, i will note that the symmetrical action of copying is still broken, and so is navigation. text editing still feels horrible, because my muscle memory is entirely tied to my keyboard actions. BUT at least now i am able to do everything. previously it was impossible to paste from the clipboard history. now everything is possible but just finicky because i need to tap on the text input for everything except for pasting.
  • 1 vote
    works today on fluxfur.com :3
  • 1 vote
    Fix is live on client. Can now log in to fluxer.com, on Version 2026.11.100, Beta 4, Channel Beta, Build 3007, Android 17 (arm64), Device Google Pixel 9a, Locale en-US, Push unifiedpush, Flutter 3.47.9, Dart 3.13.7
  • 2 votes
    To my understanding, this is not a feature in AOSP, but it is a very common extension that many OEMs and "custom ROMs" provide (i think including LineageOS? very unsure though). an example of an Android that doesn't provide this is GrapheneOS. it supports "private spaces", which is a separate OS user nested within your main one, that can be locked behind additional authentication. But it is not quite the same as the basic app locking functionality. Instead, apps like Bitwarden, Steam, Molly, Element, ente, GitHub, Fastmail, and more, all implement this "client-side". In a lot of these cases, the lock actually wraps an account, and technically don't prevent you from using the app (even though the app might require an account always). For example, if you fail the Bitwarden biometrics you can provide your bitwarden master password instead (or log out, if you feel like it). the device PIN (which a sneaky onlooker might have glanced more easily) does not grant access to the app; only the app's password or specifically biometrics that the app authorized. if an onlooker knows the device PIN and adds a fingerprint, all of the app locks i mentioned will reject the new fingerprint until it is authorized by the app's preferred password. Also, notably, besides providing different semantics from what the operating system can provide (in the case where the app lock wraps everything useful within the app), doing it "client-side" is also more flexible because you can have certain accounts locked and others not. For instance, you might have a "more private" Fluxer account that you REALLY don't want to accidentally open in public, or want to prevent a nosy family member from looking at. It would make sense to lock just that account, instead of the whole app. You could of course achieve a very similar result by using the "private spaces" (not universally supported in all Androids!), but that would also mean the private account can't receive push notifications when that space is locked (which may be preferable! but it's also annoying if that's not what you'd prefer!).
  • 4 votes
    I'm strongly against this. I think deciding the line on what is "the real bot" can genuinely get difficult at scale (though the same can be said for communities; which already have a verification mark given out by instance operators). Fluxer probably shouldn't have any kind of verified mark on bots, unless it is a system bot (operated by this instance's admins). Instead, I'd much rather like to see the ability to add connections to bots. Then, you could see that it's definitely operated by someone who controls the official help site for that bot. Moreover, this mark looks the same across instances, so it's easy to tell that e.g. "Fluxcord on FluxFur" is genuinely operated by the Fluxcord developers, instead of being just something somebody hosted early enough and got verified that way for being "the main Fluxcord bot on this Fluxer instance". With web-standard rel=me connections, you could also add a connection proving that the bot represents a particular git repo (because the invite link is in the README with rel=me), or that it really belongs to a specific public account on a microblogging site providing announcements.
  • 1 vote
    Is this not kind of technically already possible by configuring a HTTPS_PROXY for the link unfurler? have a proxy that TRICKS IT into thinking Nitter lives at x.com and Invidious lives at youtube.com. but yes, I'd love to see this kind of thing more natively supported with less jank.
  • 4 votes
    Although i agree that Twemoji is not the pinnacle of emoji, i think it's good to have a consistent emoji font, which helps messages render consistently across all platforms. Emoji fonts can disagree on the specifics of Unicode emoji, which can genuinely cause miscommunication. Therefore, I'm loosely against the idea of letting a user customize this in the official clients (but i'm not willing to die on this hill). Instead, maybe a user could choose what emoji font they wish to use for their messages? That way, it would look consistent for anyone viewing that message. But then an entire channel would perhaps look like a discordant mess of jumbled design languages. Maybe a guild could choose which emoji font it prefers? Maybe a self-hosted instance could choose its emoji font?
  • 2 votes
    wait what? 2FA is not disabled after resetting a password? that feels like it defeats the point of letting you reset your password. why even present the option if you can still be locked out after successfully doing that.
    great idea to add email 2FA
    that wasn't what i suggested! but yeah, allowing a user to enforce email approval per login (i.e. email "2"FA) would be pretty neat. this can perhaps be more secure than a weak password, especially considering that an email usually has authority to reset a password anyways.
  • 1 vote
    despite not feeling like constructing this edge case to file a real bug report for it, i will note that the "obvious" solution to me seems to only have fluxer.com as the login domain. for some users that would mean logging them out if they didn't already migrate smoothly automatically, but that feels preferable to being unable to log in, and i feel like it might drastically simplify the code. the legacy origin should probably only host the passkey bridge and never present a full login page? (maybe have that automagical migration dance on the login page, to prevent it from feeling like you got logged out. crucially, never allow you to enter credentials on the legacy domain unless it's specifically the passkey bridge)
  • 1 vote
    This is fixed for me now, with https://github.com/fluxerapp/fluxer/pull/3809. My phone web browser, which "didn't actually migrate" (but whose sessions expired) now redirects to the fluxer.com origin, as expected. I can log into the feedback site again, without manually editing the URL. based on the implementation details I would suspect that there is still a problem if the old session is not expired and you want to log into feedback.fluxer.com using a different account that has a passkey... but i don't have two accounts to test with and i don't know how to successfully construct a "legacy origin" browser state to use when reporting such an edge case, and as such i'm not actually confident it would break. at any rate, i don't feel like constructing this edge case on purpose.
  • 1 vote
    i like the implementation of just a line of text that specifically contains "Thank you for supporting Fluxer!". this is much clearer and less concerning than the stuff i initially imagine when i hear "Operator mark".
    "The admin of this instance has an Operator Pass."
    i also like the distinction of using "admin" to refer to the operator of the instance. this reduces the operator overloading. i think it's probably important that all official marketing material / formal communication should use "admin", and never "operator", lest it be confused with the "operator pass".
    we won't call it "donator", because the pass is a purchase, not a donation.
    fair enough! i was just spitballing what other terminology could be more unambiguously "recognition, not endorsement". i think "Thank you for supporting Fluxer!" achieves that goal perfectly, while causing less of a headache for the accountant.
  • 1 vote
    On my phone (which notably has not gone through the "seamless automated migration" shown in the beginning of the video i recorded for desktop (and it hasn't gone through this because i don't use fluxer in the browser on my phone)), i had the exact same experience. I still cannot log into the feedback site, because it does not let me use a fluxer.com passkey without manually fiddling about with the URL. It presents me with the same web.fluxer.app login page with the same "session expired" notice on my account. In a fresh browser, it does present me with a fluxer.com login. There's no issue when signing in under a private browsing session. notice that in this video of the issue occurring in mobile i had to manually substitute the origin twice(!) . that's odd! (i do not believe the desktop/mobile distinction is relevant, those are simply the form factors of two devices where i experienced this issue)
  • 1 vote
    as far as i can tell, bitwarden does support it (as of shockingly recently! well the implementation was there but it was intentionally disabled until recently) i think the bridge as a fallback makes sense in case a password manager or browser doesn't support related origins. but maybe that could be used in addition to RORs, that way some users could experience a smoother legacy-passkey-origin login flow that doesn't involve bouncing the user between origins each time?
  • 3 votes
    SMS verification is insecure due to SIM hijacking (a specific kind of social engineering attack, which email is less susceptible to) and because SMS can happen over insecure protocols that are vulnerable to MITM (a technical attack, which email is also less susceptible to). Why do you want SMS to be an acceptable fallback for a lost authenticator? Can't you just reset your password through email? That's the de-facto standard way to recover account login credentials on every web service. Or save your recovery codes somewhere? Or just don't use 2FA if you think you will lose your authenticator and don't want 2FA to be mandatory? the point of an authenticator is that if you lose it, you get locked out. don't use it if you don't want that behavior. SMS is expensive to send, so it's impractical to implement. Emails are free to send. Phone numbers are easy to lose access to; people change their number all the time. Email addresses, less so. Phone numbers are also more often tied to a legal identity than an email address. Collecting a phone number is in many cases functionally equivalent to collecting an ID. This makes SMS verification a notable privacy concern.
  • 1 vote
    in the video i recorded for this, i specifically made a point to show that i cancelled the prompt (and commented that i could log in otherwise) because Yes it is very unintuitive that the normal pre-passkey-migration flow requires cancelling once! i hope that can be improved indeed! (NEONSY mentioned "WebAuthn Related Origins", which perhaps sounds less hacky than the current passkey bridge? and could work in both directions(?) (or always log in on fluxer.com, even for OAuth)) (note that i realize the comment i'm currently replying to didn't have that context because the issues were merged)
  • 1 vote
    duplicate of #4574 ?
  • 1 vote
    Does Fluxer even need to do anything here? I feel like there are performance tradeoffs to this, and the S3 backing storage for media uploads is perfectly capable of implementing this on behalf of a Fluxer instance. as far as i can tell, SeaweedFS, which ships by default in the self-hosted compose recipes, does not in fact deduplicate files, but for example Garage does. It seems natural that an instance operator can choose an S3 implementation according to their needs.
    when I send that silly cat gif from my saved media ten times in a row
    it also feels natural to me that this should not cause a reupload, but just link to the same instance of the file each time. but that doesn't need global deduplication; this just needs the client not to reupload stuff unnecessarily.
  • 1 vote
    also note that logging in normally does work. my passkey did migrate successfully, it does work when i try to log into the web app directly, without initiating the login from the feedback site. (i had tested this by logging out and back in, which i did successfully on fluxer.com, and perhaps that's why my feedback site session was invalidated, leading me to discover this)
  • 1 vote
    technically i think it is the other way around? @fluxer@social.fluxer.com is the intrinsic handle and @fluxer@fluxer.com is the preferred/canonical alias. i simply copied the handle from my fedi software which doesn't properly support aliases, so it only shows the intrinsic handle. 😔
  • 1 vote
  • 3 votes
    i would LOVE to toss the cat around
  • 1 vote
    Can reproduce on the GrapheneOS stock launcher. Not launcher-specific.