element-web
element-web copied to clipboard
element-hq
@room should require a direct, intentional, action to ping
Description
This is a tweet brought into matrix through a bridge. That should not ping me.
Version information
- Platform: web (in-browser)
- Browser: Chrome 62
- OS: Windows 10
- URL: riot.im/develop
I was about to 100% agree, then I wondered - what metric would we use to decide that the twitter bridge can't use @room notifs? It's a bridge - so are IRC and Slack, and I can certainly see supporting @room notifs from those being valuable. Even if it were a bot, I can still see @room being useful (though open to abuse, certainly).
The reason we'd be happy to exclude all @room notifis from twitter is that we no nobody on twitter is trying to get our attention - it's a one-way broadcast and the @room paradigm doesn't map to twitter at all. So maybe this should be something that the bridge impl takes care of - specifically agreeing with the homeserver that its messages won't be processed for @room notifs (and whatever else makes sense here).
There might already be support for this in synapse/matrix fwiw; I'm not intimately familiar with the details of bridging.
I can imagine a situation where you want some existing software, that only has Twitter support, to alert everyone in the room because a message is important.
Is this not actually a problem of non-admins being able to alert the entire room? Or is there a reason for the RSS feed to be an admin in the Matrix room?
The spec for @room should include some way to indicate on a message that it is intentional, to avoid accidental and unwelcome pings. Something like { body: "@room: Hello world!", mentions: ['@room'] } in the m.room.message. This could potentially also be expanded to user mentions.
Clients would probably end up just doing if (message.contains('@room')) content.mentions = ['@room'];. Bridges could handle this on an individual basis (where the Twitter bridge would probably leave mentions empty/omitted).
worth noting that other platforms like Slack warn you before you do this, which would be a good way to stop most of the cases. Bridging through Twitter is a whole other story, but maybe we can stop people from accidentally sending @room (either in a reply or in a message).
