!DeRvyHqkqkIBbBtwsO:matrix.org

Homeserver Developers

377 Members
If you are building a homeserver, or you want to talk to other people that build homeservers, then this is the place for you! πŸ₯³143 Servers

Load older messages


SenderMessageTime
7 Sep 2024
@coolbeans:skid.lifecoolbeans joined the room.02:18:44
@coolbeans:skid.lifecoolbeansπŸ‘‹02:19:57
9 Sep 2024
@dizzy97:matrix.orgdizzy97 joined the room.11:44:48
11 Sep 2024
@daedric:aguiarvieira.ptRicardo Duarte changed their profile picture.18:09:28
12 Sep 2024
@abuse:matrix.orgAdministrator banned @ambiguous_yelp:matrix.org@ambiguous_yelp:matrix.org (spam).22:20:43
18 Sep 2024
@joshsimmons:matrix.orgJosh Simmons (m.org) changed their display name from Josh Simmons (m.org alt/no DMs) to Josh Simmons (m.org).07:52:31
@joshsimmons:matrix.orgJosh Simmons (m.org) changed their profile picture.12:29:12
@joshsimmons:matrix.orgJosh Simmons (m.org) changed their profile picture.12:32:16
@bkil:matrix.orgbkil To HS developers. If you want to test your HS implementation with a room created with a malformed create event, please join #Office of the Matrix.org Foundation . HS other than Synapse seems to have a problem with it, so it may be a good idea to open a ticket in the respective repository. 22:05:19
19 Sep 2024
@cat:feline.supportCat Malformed predecessor mapping specifically 06:20:45
@cat:feline.supportCat So it’s an end user supplied bit of data. 06:21:03
@ugp:matrix.org@ugp:matrix.org left the room.06:24:27
@s7evink:matrix.orgTill
In reply to @bkil:matrix.org
To HS developers. If you want to test your HS implementation with a room created with a malformed create event, please join #Office of the Matrix.org Foundation . HS other than Synapse seems to have a problem with it, so it may be a good idea to open a ticket in the respective repository.
No..
07:32:09
@s7evink:matrix.orgTillThe room needs to be fixed07:32:20
@s7evink:matrix.orgTillhttps://spec.matrix.org/latest/client-server-api/#mroomcreate07:32:47
@cat:feline.supportCatTheres enough broken rooms out in the wild.07:59:25
@cat:feline.supportCatThat some level of handling is required. 07:59:52
@cat:feline.supportCatI know rooms that have no event ID linked but only a precursor. And then we have this new one.08:00:15
@cat:feline.supportCatAfter all its a end user supplied field that lacks validation for years in Synapse.08:00:30
@s7evink:matrix.orgTillAnd there always will be broken rooms, if we don't enforce validity of events. So, my stance is, neither Dendrite nor Conduit (and forks) are doing something wrong by rejecting the create event.08:10:36
@cat:feline.supportCatTbh the fix for this long term is have Synapse enforce Required fields.08:21:36
@neilalexander:dendrite.matrix.orgneilalexanderSynapse should never have accepted the event to begin with. Regardless of whether it is optional or not, the field is known to the spec as having a specific type when it is present 08:47:37
@cat:feline.supportCat
In reply to @neilalexander:dendrite.matrix.org
Synapse should never have accepted the event to begin with. Regardless of whether it is optional or not, the field is known to the spec as having a specific type when it is present
it has required properties if present so yes i agree Synapse should enforce that.
08:48:48
@neilalexander:dendrite.matrix.orgneilalexanderI'm really tired of having to beat the "interoperability only works if we actually play by the same rules" argument but it applies here yet again, and it's mostly unfortunate/ironic that it's the Foundation's poster room that is broken because of an interop issue when the Foundation are supposed to be the ones driving interoperability08:51:18
@neilalexander:dendrite.matrix.orgneilalexander * I'm really tired of having to beat the "interoperability only works if we actually play by the same rules" drum but it applies here yet again, and it's mostly unfortunate/ironic that it's the Foundation's poster room that is broken because of an interop issue when the Foundation are supposed to be the ones driving interoperability08:51:47
@neilalexander:dendrite.matrix.orgneilalexanderSo IMO neither Dendrite nor Conduit are "broken" here and do not need to be fixed08:52:34
@travis:t2l.ioTravisRThere's multiple issues here, which don't make a single project at fault. I don't think it's beneficial to create sides here.08:56:57
@neilalexander:dendrite.matrix.orgneilalexanderTo be clear, I understand human error is a thing and it's ultimately a mistake β€” they do happen β€” but I just don't think it's useful to say that this is a good litmus test for maintainers and then expect this to be fixed in implementations that are actually working properly. Otherwise this is just going to become one of those "de facto" bugs that everyone implements and then there's no guard for the next time it happens08:59:03
@travis:t2l.ioTravisRThe bug is that this is being treated as an authorization failure when it's a validation failure. The server should (at best) be redacting it and serving that to the client, not failing to join. Meanwhile, client-server APIs should be preventing invalid data from getting through.09:01:25
@travis:t2l.ioTravisRWe can also improve the spec to avoid the situation a bit more clearly in the future.09:01:43

Show newer messages


Back to Room ListRoom Version: 5