Damn has it been that long already? I saw the moment when it broke, it appeared to coincide with a network outage after a couple weeks of being reliable. It's got to be some kind of simple little bug so I even started looking at the code but didn't get anywhere yet.
I block all mags from all Cloudflare instances. So sometime after posting to homeimprovement@lemmy.world I blocked it. I still get notifications (which is good), then I unblock it just to read the response then re-block. That’s not likely related, but something happened in the past couple days that apparently caused a “seen” flag to get unset. It’s not in itself worth chasing up but maybe it’s a clue as to whatever happened.
This might be the same issue that brought kbin.social to its knees, only fedia/mbin is giving up instead of the infinite retries which are crippling the other platform.
One comment in one of the kbin threads of a guy who hosts his own forked instance said it's probably caused by a Lemmy update, which I guess makes sense. kbin and mbin aren't the same code wise (can't even use the nsfw unblur here), and kbin is also a version ahead of mbin so they might just handle the same issue differently. A lot of the magazines / communities I checked have their latest threads from like a week ago when viewed through fedia. kbin ironically has similar examples, but different ones to fedia. Whatever the cause, it obviously shines a negative light on any 'bin instance and to a degree onto the fediverse. For me, I would really like to avoid having to deal with the shitty Lemmy UI and support the Lemmy devs directly.
I wouldn't exactly say mbin is behind kbin, as things have diverged much since the fork. For example, mbin has moderators from lemmy, and lemmy shows mbin moderators. As of latest, mbin gets moderator reports from lemmy (and maybe the other way, I didn't closely review that PR), things like sensitive groups are marked correctly, subscriber and follower counts come from source... Still, there's a lot of work to do for the user experience, many people have made that clear lately, so I'm hoping mbin will be able to gain the contributors needed to make the changes people want to see
Also, yea a lot of /kbin tweaks won't work. I do want to make an eye toggle button to make visible all nsfw on a page, but that might be hindered a bit due to infinite scroll (as well as a user setting to just have it shown by default). In the meantime though, it's just the structure of data changing from javascript to css, so it's always possible to do. I wrote a tampermonkey script able to do so as an example:
@jerry They range from 2.7 to 9.1 MB, though looking at the ones that did post successfully, they seem to be less than 5 MB. I think I ran into a size limit.
As the diff mentions the default max size is 6MB and 5s upload timeout. While php.ini declares 10MB defaults. It's a bit weird they don't match, but there's probably followup on 1. allowing instance owners to configure the size of images they are ok with hosting 2. maybe more research into what other platforms allow, there's a lot of discussion about thumbnails and the sort since remote instances cache things themselves, and broken thumbnails make things look like they aren't working across the fediverse
Was there any data loss? The comments tab of my profile only shows 11 comments which only go back 1 week (perhaps to around this “we’re back” thread). I should have much more than that.
As part of the database clean up, I had to remove many corrupt records, so the answer is yes there are missing posts and comments, however I do have backups have I have been attempting to find ways to reinsert them without recreating the problems. I’ve not had much luck with that yet.
...and it's been fixed!
Credit to Melroy and asdfzdfj for picking it up and coding the fix in what ...7 hours?
For the wider audience, this is now fixed in the mbin code, it won't resolve here until next time @jerry updates fedia.io which could be a few days, or even week or so depending what other changes they've made and are making.
I did not update last night - there are many in flight code changes and I’d prefer not to cause more problems, so letting that settle down for a bit before I update.
Ended up plagued with this again for a few days, and figured it was server-side issues.. but eventually thought "why not check in another browser", so I did, and the problem was gone.
I don't know what the problem was and if it's due to kbin, the swap to mbin, is fedia-specific, or browser-specific, or some-action-that-I-took-specific, but clearing all the cookies and data in my browser for fedia.io has put things back to sane. If you read this and are having similar issues, consider trying as much (remember doing this will log you out).
Officially not.. but the development slowed down too much and was too restricted by Ernest. I wanted to avoid a fork. But I didn't saw any good alternative.
Is Ernest OK? Maybe the explosion of interest was a bit too much to handle right now. Out of curiosity, why did you fork it to github and not stay on codeberg?
I don't know what is happening with Ernest, he said there were families issues. He did respond to me on October 3 for the last time. However, developers were NOT allowed to merge pull requests from others. He stopped developers from merging code. He couldn't let it go, which is a problem if you are not in for weeks or sometimes months. The issue was that development become to a halt, contributors were no longer motivated! I tried to discuss this topic with Ernest multiple times now, without any answers. At some point it was the final straw. I forked the project and introduce a C4 Wow based on trust, allowing dozens of people to have owner rights and giving back the control to the developers, contributors and users or admins.
Moving to GitHub was only done because Codeberg was down too often in the past year. Which was very frustrating when you want to work together with people. So I also moved to GitHub with GitHub Actions during the forking.
I was also trying to prevent a fork, but I didn't saw any way out. Hence the fork by the community, for the community. I hope so as well, the idea is that we work as a real team and active contributors have GitHub owner rights. We peer-review each other code and are allowed to merge pull requests. There is no single maintainer, we are all maintainers.
Thank you for all your hard work. You and @melroy haven't given up on this community, and I'll stand by you too. I embrace the jank and appreciate the fine communities we've made here.
Is this a Cloudflare problem? Without any error messages this may as well just be a Fedia bug. Every other Kbin user seems to be posting comments just fine.
the client is javascript after all
How edgy! It’s written in Typescript, though.
Best advice I can give you is to open up the dev tools and look for the exact error message that Kbin is hiding from you for some reason. Also do the usual checks (disable adblockers/all addons, try another browser, etc) for browser related issues.
Though there’s another odd behavior: my profile shows all my own posts except the posts in that #lemmyWorld thread. But I do get notifcations if someone replies to me in that thread.
Every other Kbin user seems to be posting comments just fine.
I don’t see how you could know that. If someone is blocked, there’s no metric AFAIK.
I’ve retried several times, even with a long delay. I also tried re-entering the thread (which is often the fix for kbin-specific failures) and that did not work either.
(edit)
Note that to send this reply to you here in this pure kbin mag I had to submit twice. But in that #LemmyWorld even many retries makes no difference. I would love to have a text-based client about now.
The same person who runs Fedia.io also runs infosec.pub, so you could try using neonmodem with that Lemmy server without handing your data over to a new person.
fedia
Top
This magazine is from a federated server and may be incomplete. Browse more on the original instance.