115 Members
Discussion of specific Matrix Spec Change proposals | https://matrix.org/docs/spec/proposals | Design draft folder at: https://drive.google.com/drive/folders/0B4wHq8qP86r2ck15MHEwMmlNVUk 77 Servers

Load older messages

2 Jul 2020
3 Jul 2020
@reivilibre:librepush.netreivilibre Anyone know off the top of their head whether an event's content field is always a dict if it's provided? 08:49:08
Download image.png
@timo:koesters.xyzTimoSpec calls it object which JSON's name for dicts afaik08:51:39
@reivilibre:librepush.netreivilibreThanks, perfect. I was searching for 'EventContent' as it's called in other parts of the spec, oups08:52:16
@timo:koesters.xyzTimo * Spec calls it object which JSON's name for dicts afaik08:52:15
@reivilibre:librepush.netreivilibreby the way, I believe you are responsible for Conduit, which makes me happy inside, so thanks for doing that :)08:56:15
In reply to @reivilibre:librepush.net
Anyone know off the top of their head whether an event's content field is always a dict if it's provided?
as a rule: applications need to treat event data as untrusted
@richvdh:sw1v.orgrichvdh ie, they should not segfault if they get a content which turns out to be an int 08:57:04
@timo:koesters.xyzTimoYeah, not segfault, but it's okay to reject, right?08:57:28
@richvdh:sw1v.orgrichvdhas a client: I'd have thought you should just ignore it08:57:48
@reivilibre:librepush.netreivilibreyes true (and somewhat 'of course', but if we complain and don't handle it in an innocent way I think that'd be OK?08:57:57
* @timo:koesters.xyzTimo doesn't need to worry about segfaults, he uses rust08:58:00
@reivilibre:librepush.netreivilibreI've made Python segfault doing semi-innocent stuff before now o.O08:58:57
@richvdh:sw1v.orgrichvdh(sending malformed events is a good way to make everyone's riot-web stop working. so yeah, not segfault, but...)08:59:42
@julian:foad.me.ukJulianFRedacted or Malformed Event09:39:10
@julian:foad.me.ukJulianF * 09:40:22

Is this a spec bug? actions should probably not be [enum] as each action could have parameters at which point it should be a dict e.g.

    "set_tweak": "sound",
    "value": "bing!"
@reivilibre:librepush.netreivilibrein: https://matrix.org/docs/spec/client_server/r0.6.1#put-matrix-client-r0-pushrules-scope-kind-ruleid-actions13:14:31
@reivilibre:librepush.netreivilibrewelp, I jjust misclicked 'Search' instead of 'Copy' when copying my access token to test something :/13:22:57
In reply to @reivilibre:librepush.net
in: https://matrix.org/docs/spec/client_server/r0.6.1#put-matrix-client-r0-pushrules-scope-kind-ruleid-actions

I think this is probably a spec bug; how to handle it?

Synapse implements what I would consider 'what should have been specced', i.e. returns

    "actions": [
            "set_tweak": "highlight",
            "value": true

even though the set_highlight is not a string...

@andrewm:amorgan.xyzanoaIf it's one or two instances that are typos, then file a bug at https://github.com/matrix-org/matrix-doc/issues and/or feel free to submit a PR with a fix. But... if it's intended and Synapse acts a different way, an MSC is in order.14:05:31
@reivilibre:librepush.netreivilibreI think it's a typo because it makes no sense otherwise (you can't properly represent an action as a string or enum)14:06:23
@david:vovo.id.audavoI think the examples also have set_tweak objects as well14:07:09
@andrewm:amorgan.xyzanoaProbably a spec bug then and you can submit a PR14:46:19
@freenode_MichaelRaskin:matrix.orgMichaelRaskin joined the room.19:51:44
@sbug:1bit.sesbug changed their profile picture.22:48:48

There are no newer messages yet.

Back to Room List