You can use pg_stat_statements to find slow queries. Try sorting by top total_exec_time.

Good example of AI making stuff up instead of simply saying “I don’t know”.

The key is refreshed after 24 hours so it will work if you wait a bit.

There were optimizations related to database triggers, these are probably responsible for the speedup.

https://github.com/LemmyNet/lemmy/pull/4696

Alright Ive added @[email protected], @[email protected] and @[email protected] as mods and removed the inactive ones.

I just opened an issue about it: https://github.com/LemmyNet/lemmy/issues/4744

This, particularly reports are not fully federated.

True there are a lot of issues related to notifications. I will see what I can do.

Here is the relevant issue, but no one is working on it currently.

None of that matters if Mastodon doesnt implement these suggestions or standards. And from past experience its extremely unlikely that they will. Thats why I think its best to ignore what Mastodon does, its not our concern how they decide to render things.

Well written, it would deserve a separate post.

One of the comments mentions that another app can trigger search through an Android intent. So its better to be safe and close any potential vulnerabilities, but this doesnt seem particularly useful for an attacker.

I dont have time to read all that. The problem with Beehaw is that the admins are extremely entitled, as if we had some obligation to work for them for free. Similar to what is described in OP.

However we are consistently improving the mod tools, and accept contributions in that area. You can see in the dev updates.

What is this “stance on mod features” that you are talking about?

Im a former contributor to F-Droid with various merged pull requests. Looking at the indicated pull request I really doubt that it was an intentional attack. First of all its easy to forget for a new developer to escape SQL parameters, and the docs dont even mention a risk of SQL injection attacks. And of the users pushing for the PR to be merged, one is a long-time F-Droid contributor, and the other also looks like a real human with many contributions in other repos, so no sockpuppets in sight.

It simply looks like standard open source behaviour, for better or for worse. A new user makes a contribution for a highly demanded feature, and users want it to get merged as soon as possible. Maintainers are discussing the big picture of the change and want to avoid breaking changes, without getting into code review yet. The new contributor seems unwilling to make any design changes to his PR, and gets frustrated that it doesnt get merged as is. The potential vulnerability is only noticed half a year after the PR was opened, at which point it was already de facto abandoned. So not an attack, but simply a developer who is new to open source and doesnt understand how the process works.

We applied for funding last August, but unfortunately we are still waiting for it to be finalized. Seems like NLnet is quite overloaded these days.

We didnt make any changes to the Lemmy version running on this instance during the past week. So it must be something else…

The problem is that a server could very easily lie and claim to have captchas when it really doesnt.

Someone on in the Matrix chat tried this and didnt have any problems.

I see now, if an instance has any site languages configured those will be applied for new users. You can see it in /api/v3/site field discussion_languages. However both lemmy.world and lemm.ee return all languages there.

Edit: Im removing this as part of the PR to set new user languages from accept-language header, it doesnt make sense anymore with that.