The new major version of Lemmy is now ready, and we need your help with testing. Most importantly it uses HTTP for API requests now, which is much more efficient than websocket. Additionally Two-factor-auth is supported. There are also countless other improvements and bug fixes.
You can register on any of the following servers to start testing, no approval required. You can post to your hearts content to find out if anything is broken. The test instances only federate with each other to avoid affecting production instances with spam.
If you encounter any bugs that aren’t present in 0.17, open an issue and mention in the title that it happened with a release candicate version. Over the next days we will publish new RC versions to fix bugs that will invariably pop up.
Instance admins can try the new version by using Docker images dessalines/lemmy-ui:0.18.0-rc.2 and dessalines/lemmy:0.18.0-rc.1. Make sure that working backups are in place. For production instances its better to wait at least some days for the major issues to be fixed.
Peeps, I am seeing some really worrying trends on https://lemmy.fediverse.observer/list. Many instances are quickly filling up with thousands of spam accounts which will soon be unleashed on the threadiverse. While bots can bypass captchas, they at least limit the simplest scripts. We are going to face this really really soon https://lemmy.dbzer0.com/post/87753
Can we at least add support for disabling VPNs, or using some other captcha solution like recaptcha? IP rate limiting is useless with VPNs, and email verification is more trivially bypassed than the current captchas.
I like the ideas of good captchas or text applications to join. However, using one’s IP kinda goes against the idea of privacy. I’d prefer if we find alternatives.
One alternative that already exists and has been working well for instances that use it is an application process.
The server can see your IP when you connect to it and IPs are not sensitive either way. That’s not a privacy issue.
I’ll paste my comment to @muddybulldog@mylemmy.win, which also applies in this situation: I see your point. What if I use VPNs with a killswitch? —meaning that I can only ever connect to the internet through my VPN. What if someone is avoiding surveillance from their government? Should they disable their killswitch and risk them finding out they’re part of something ‘political’ like Lemmy?
Using an IP in this way has no impact on privacy. Instances already have your IP info as a result of interacting with them.
I see your point. What if I use VPNs with a killswitch? —meaning that I can only ever connect to the internet through my VPN. What if someone is avoiding surveillance from their government? Should they disable their killswitch and risk them finding out they’re part of something ‘political’ like Lemmy?
I would also imagine some tooling to make it easier to remove spam accounts might be helpful, especially if they start acting up.
Additionally, even once spam bot users have been banned or bounced by failed email verification or whatever, they continue toward the user count of the instance. Not really a functional problem, but the growth of lemmy is garnering some attention, and with the bot account explosion, this growth looks astounding.
It may become a rather bad look once lemmy’s user count is basically seen to be all spam bots.
So maybe some way to adjust user counts? I’ve seen something like this in the GitHub issues I think.
How can you tell they’re spam accounts btw?
Small instance, open signups, rapidly growing users. On balance, given the issues others are having, it’s probably bots creating the accounts.
On my instance if you looked in the database they all had gmail emails and all had the same pattern to the email. We were adding 20 users a day, then suddenly had 100 new accounts in an hour. There was a lot of talk from other instance admins seeing the same thing.
On top of that there’s also activity, I saw a server with only 7 posts for 6K users, there’s no way those are real people.
A bit off topic, but can something be done about the power mods? I see a few users already forking every subreddit trying to ensure they remain a mod. No user can meaningfully manage 50-100+ communities.
Please consider capping the limit to 20 or less. First-mover advantage is huge, so starting up a community down the road to prevent this consolidation of privileges is likely out of the question.
If you get a bad mod, you can always move to a community on another instance. That’s one of the advantages of federation.
I think this is something reddit users generally have a hard time grasping about lemmy, including myself.
One of the fundamentals of the fediverse is that there will be communities with the same name on different instances. Users can subscribe to good ones and / unsubscribe from bad ones as they wish.
I do see a pure numbers challenge. We need more features (maybe on clients, or maybe in the Federation) to see all subbed instances of
c/newstogether except whne it’s important we don’t.
And an advantage of community redundancy. Old quiet communities can be literally brought back to life if we enough people decide to move.
I think you have the right problem with the wrong answer.
It’d be better if communities could subsume other, worse-moderated communities with the same name in some integrated/organic way.
I mean, I could run c/politics on some server, but if another 20 or 30 instances agree on c/politics that’s the winner. If they agree on c/politics because it’s the better one and cross-moderates in some way, more power to them.
Name and shame.
Report from a tester that “Sign Up” with duplicate username or email is producing JSON errors and not a proper error message: https://github.com/LemmyNet/lemmy-ui/issues/1356
Trying to register a pre-existing username/password still doesn’t return a proper correct error, but at least it returns an error instead of spinning endlessly
The fix for that isnt merged yet, will be in one of the next RCs
0.18.0-rc.4is now pushed up.Congratulations on the improvements to the protocol and to you 2 leftist sexy pimps for keeping the FOSS philosophy alive.
On behalf of the Mujican community, I send you a fraternal hug and we reiterate that we are with you in whatever is needed, wherever the fediverse takes us.
Can we please get the 10K char post limit raised to 50K or even 100K chars in v0.18? Many mods are desperate to get their wikis/guides ported from reddit and this has been a showstopping issue.
This is a UI issue. As a temporary solution, you can modify the post (maxlength) using your browser’s dev tools until lemmy devs implement a fix.
We’ve pushed up a new UI version for testing there:
lemmy-ui:0.18.0-rc.3.Back-end should also be done in a few hours.
Navigating to a different page scrolls all the way to the bottom of the page. I assume it’s part of some feature to remember the previous position (i.e., when navigating back), but I don’t think it should apply when using the pagination navigation. I’m on mobile.
Thank you for your work :)
p.s. test instance names are fantastic! :D
Is there a full list somewhere of all new features for us to test?
Only ❤️ for you guys!
omg yiss!
Posted this on Martrix… but receiving this error on all three instances on clicking the “Sign Up” button after entering my details:
edit: Also DS seems to require admin approval for new signups.Sorry, we merged someone’s bad PR without testing, we’ll revert it and start another build.
FWIW I tried an hour later and it worked. Not sure if something was fixed in the back end.
Appending the char “!” to the password gives the error “SyntaxError: JSON parse: unexpected character at line 1 column 1 of the JSON data” - same screenshots posted by @Lodion - that’s on Enterprise (haven’t tried other instances yet).
Removing the ! let me create my user just fine.
Edit: oh just saw mod message, will wait for next build then try again.
Oh that kinda seems like a bad sign regarding proper data validation. Hopefully it’s just a small bug and not part of a larger issue.
