| 24 May 2025 |
 uep | so i expect the restart has probably cleared it and we won't know | 09:01:50 |
 K900 | You have to !mjolnir unban @mx:id true | 09:01:58 |
| K900 | The true is what removes from the redact list | 09:02:10 |
 hexa | I hate that they did htat | 09:02:17 |
| hexa | * I hate that they did it like that | 09:02:21 |
| uep | until it happens again. I can increase the threshold a little more, because we get false positives for things like code blocks with @@ patch lines | 09:02:30 |
| hexa | ultimately we cannot automate protections against everything | 09:02:57 |
| hexa | if at least the imagery is gone that is the big step | 09:03:18 |
| hexa | and I still wish that was a per room setting | 09:03:37 |
| uep | no. but the mentions spam protection is especially disappointing because the match is so dumb (it will literally fire on N consecutive @'s) and the spammers actually do spam mentions and so it's a good trigger | 09:04:06 |
| hexa | https://the-draupnir-project.github.io/draupnir-documentation/protections/mention-limit-protection | 09:05:38 |
| hexa | reads a bit more nuanced | 09:06:13 |
| hexa | but surely not a silver bullet | 09:06:39 |
| uep | yeah that seems better, though it will initiate a ban so I hope it's also not as prone to false positives (and that's the most useful change if so) | 09:07:25 |
| hexa | frankly, if someone wants draupnir someone needs to pick up the PR that ships the module and test | 09:08:15 |
| uep | anyway, draupnir is allegedly a drop-in replacement and can also be swapped back if for some reason you don't like it, so i say we jusxt try it | 09:08:27 |
| uep | there's some extra optional stuff but that's optimisation and not required. | 09:09:05 |
| hexa | https://github.com/NixOS/nixpkgs/pull/400194 | 09:09:30 |
