- cross-posted to:
- foss@beehaw.org
- technology@beehaw.org
- cross-posted to:
- foss@beehaw.org
- technology@beehaw.org
You must log in or register to comment.
@dessalines@lemmy.ml Thanks for the information here and all the hard work you have put into this release.
Gotta say tho, as the maintainer of Lemmy-Swift-Client, breaking API changes like this without an API version bump, make API development within the community incredibly difficult.
So my question to you would be, what is the purpose of having
v3in the API path, if the true test of API compatibility is the GetServerResponseversionfield? And breaking changes will occur in GetServerResponseversionchanges as opposed to the version in the API path? That doesn’t quite make sense to me.Would love your perspective so I can figure out how to best design the package API to accommodate client developers who might have to contend with multiple server versions.
The lemmy API still hasn’t hit a version 1.0, and should very much be considered beta, with a lot of active and breaking changes. When we do stabilize it, then we can start to make these breaking API changes more solid. The
v3should probably just go away at this point, because we have too much active development and API changes to justify it.What we do on lemmy-js-client, which has its types auto-generated from rust, is use tags that match our lemmy release semver version.
I’m not sure how you built lemmy-swift-client ( I hope its auto-generated from either the rust or lemmy-js-client types), but you could do the same thing: use tags to version it, then applications could use those tagged versions.
WebSockets … causing live updates to the site which many users dislike
I appreciate all the work in this release. It’s insane how much you packed into one release. Well done. I am most excited about the live updates going away. It was quite disruptive. Thanks for that.
That said, WebSockets can be implemented very efficiently. I run an open source notification service called ntfy, and the public instance ntfy.sh currently keeps 6-8k WebSocket connections and thousands more HTTP stream (long polling HTTP) open, all on a 2 core machine with 4GB of RAM. My point being that WebSockets can be implemented very efficiently. Though in Lemmy’s case it’s likely not necessary.
– Another thing I wanted to notice is that I am missing mentions of security issues in the release notes. There are some tickets that sound really really really bad, like this one: https://github.com/LemmyNet/lemmy/issues/3060
Isn’t that more important than anything else?
We’re very busy but we’ll get to it when we can.
That said, WebSockets can be implemented very efficiently.
I agree, the main issue I was actually seeing with Lemmy’s use of WebSockets was that when opening the main page it was continuously streaming all posts from the server (including posts in communities not subscribed to) to the browser client.
I laughed a lot on it 🤣
That’s a LOT of changes. Kudos to all the people involved.
wow TOTP support is always welcome!!
Is it safe to update via ansible?
Yes. We have one issue open having to do with site icons, but everything else is smooth.
Good to know. Thank you for your work!
I tried to upgrade via the instructions doing a
git pulland then running ansible again and it totally broke my site with aserver errormessage. I ended up reverting back to 0.17.4.EDIT: It looks like they added some extra NGINX proxy stuff in there. All that broke my instance and I had previously just deployed via ansible following the instructions on their page. I would stay away for now.
Yeah, I’ll better wait for now.
I really appreciate the work that you’re doing but…
Captchas are not available in this version, as they need to be reimplemented in a different way. They will be back in 0.18.1, so wait with upgrading if you rely on them.
What the hell? Another bad default where newbies will continue to launch instances attracting 10k+ bot signups per day and new users will keep getting disgusted by Lemmy if they land on one of those spammy instances. Making captchas a default should have been a priority and I’m sure no one would have complained if the 0.18 release was postponed to make that happen.
The feed desperately needed a fix, so I’m glad they pushed it out. But agreed, they should be prioritizing methods to reduce spam and bots in the next release, even if they need to push out a relatively feature-sparse version first
I was tracking the Race for Captchas in
0.18.0and I’m pretty surprised how it went in the end.
I thought captcha were re enabled on the rc For this but the notes say it’s gone again. Adding one completely stopped botted sign ups for me
Captchas have been restored by a PR shortly before release and almost immediately been removed again by reverting the PR. The PR author has been asked to re-create his PR and then
0.18.0got released without Captchas.Wow that seems… Odd
Man, i appreciate updates, but lemmy_api_common is now broken.
very nice, updated with zero issues.
are there any plan to integrate lemmybb as optional web interface ?
Not with the default installation if thats what you mean. You can already install it manually alongside lemmy-ui.
I don’t know how this works but I think I read somewhere that this version fixes the new post spam on browser, I checked and it’s still the same, when will that be fixed? Is it an instance thing?
The instance has to be running the updated version.
What does new post spam mean?
I noticed in the web client used on mobile, there would often be new posts flooding in at the top of the page. Maybe a page’s worth at a time. I’m still using web client on mobile on lemmy.ml and haven’t noticed this since the deployment of 0.18 yesterday.
