!GsmxjHfeAYLsTEQmjS:nixos.org

Matrix Meta (Nix)

619 Members
Discuss your proposals for the Matrix space here, before suggesting them in #matrix-suggestions:nixos.org176 Servers

Load older messages


SenderMessageTime
18 Jul 2025
@cat:feline.supportCatThe reason why communities with corporate influence tend to organise healthy is because corporations learned how to organise properly already. They know that Security, IT administration and Human Relations are not something that the same people are good at. So separate them. Communities need to understand that your admins need not be moderators. Leave moderation to moderators and administration to administrators even if they sometimes are the same creature with diffrent hats. So thats the TLDR of what my post will be about and this will materialise in Draupnir with Draupnir supporting this pattern of having a team own stuff so that team members can come and go without it being a security risk.17:33:21
@cat:feline.supportCatSo since Draupnir must be able to create your rooms now it also needs a manual room upgrade command even if its a rudementary one at that.17:34:24
@emilazy:matrix.orgemily
In reply to @0x4a6f:matrix.org
I did something back then: https://github.com/NixOS/infra/wiki/Tombstoning-a-room-on-Matrix
good grief has @k900:0upti.me really been doing this
17:35:22
@cat:feline.supportCatThats a creative solution to the problem.17:36:54
@cat:feline.supportCatusing 2 tombstones.17:37:03
@cat:feline.supportCathttps://gitea.blesmrt.net/mikaela/scripts/src/branch/master/bash/matrix-upgrade-room.bash is the script i like to refer to.17:38:08
@emilazy:matrix.orgemilyI'm inferring that the bug v12 tries to fix is "an attacker can make a room's state permanently screwed up beyond fixing"17:39:22
@emilazy:matrix.orgemilythough I wouldn't have expected them to put out that softening statement on the urgency if any arbitrary attacker can do that, hmm17:39:35
@f0x:pixie.townf0x huh, but the predecessor event_id being the tombstone wasn't even that important right, just happens to be what Synapse does since it can create chicken and egg at the same time 17:40:00
@cat:feline.supportCatits important but the exact event ID is not.17:41:02
@cat:feline.supportCatas in feel free to pick any event ID you want people to get jumped to17:41:20
@cat:feline.supportCatAnd even then that feature is maby getting removed for v12 because they are being dumb with thinking their chicken and egg dance is actually needed.17:41:46
@f0x:pixie.townf0xright yeah17:41:54
@cat:feline.supportCatmy powershell script when i do upgrade with notice17:42:11
@cat:feline.supportCati literally use the event ID that the API returns as the value 17:42:32
@emilazy:matrix.orgemily
In reply to @charles:computer.surgery
I personally am interested in other protocols at this point
(ones that exist or not?)
17:42:38
@0x4a6f:matrix.org[0x4A6F]Best thing is to also disallow changes to the tombstone room, otherwise nick changes and avatar changes might be annoying.17:42:42
@cat:feline.supportCatas in my notice message is the event ID i use.17:42:43
@cat:feline.supportCatThose cant be blocked but yes do send a PL event to block everything else.17:43:04
@f0x:pixie.townf0xyeah that makes sense, in the past I just used the last message event id in the old room17:43:06
@cat:feline.supportCatAnd that spam is why the removal is so dumb17:43:15
@cat:feline.supportCatbecause now we have to scroll thru it17:43:21
@cat:feline.supportCatinstead of jumping past it.17:43:26
@charles:computer.surgeryCharlesnot sure yet, i want to do some shopping around; maybe something i would be comfortable using exists already. i'm looking for something with, uh, worse availability guarantees, i guess17:44:17
@emilazy:matrix.orgemilyI hope searching the old iteration of a room seamlessly isn't going to break17:44:18
@emilazy:matrix.orgemilyexcept that's also a threat model if the room owner can continue spamming in the old one now huh17:44:36
@cat:feline.supportCatThat was always a threat model17:44:51
@cat:feline.supportCatyou cant block your own ability to send messages after the first PL event.17:45:04
@emilazy:matrix.orgemily
In reply to @charles:computer.surgery
not sure yet, i want to do some shopping around; maybe something i would be comfortable using exists already. i'm looking for something with, uh, worse availability guarantees, i guess
I love to sacrifice availability. it's my favourite thing to do
17:45:13
@cat:feline.supportCatThats the only time you can block your own ability to send an event in v11 and before.17:45:16

Show newer messages


Back to Room ListRoom Version: 6