07:16 <mathju> spacekookie: I added an RPC tests with which you can trace the panicking error when creating a chat-room

07:16 <mathju> it's in this merge request: https://git.open-communication.net/qaul/qaul.net/-/merge_requests/59

07:23 <mathju> to run it: cd libqaul/rpc; cargo test

12:57 <spacekookie> Oh cool! I'll check it out today

13:47 <spacekookie> mathju: so I looked at the rpc test that was failing, and you got some of the parameters mixed up

13:47 <spacekookie> I fixed it and will push that in a second

13:47 <spacekookie> But it is making me think about making that failure in ratman less...fatal. Basically the problem was that you were including the local user in the "start_chat" parameter, which is implicit. So it tried to route to a user that was local, which it can't do

15:04 <mathju> Ahh, OK, this was the bug ... thanks!

15:05 <spacekookie> Now, the chat service still doesn't properly work, and I'm not really sure _why_. It seems like the receiving side of the chat message is never notified of new messages

15:05 <spacekookie> Even though the subscription feature does in general work

15:07 <mathju> is there a way to see the routing table?

15:11 <spacekookie> Sort of

15:12 <mathju> it would be interesting to get it as json, so we could visualize it easily in the browser :-)

15:12 <spacekookie> Ah, sure yea we can havu an api for it

15:12 <spacekookie> For testing just wrap the access here https://git.open-communication.net/qaul/qaul.net/-/blob/master/ratman/src/core/routes.rs#L89 in a dbg!()

15:22 <mathju> OK, thanks!

15:22 <mathju> An API would be great :-)

15:23 <mathju> I guess local "routing" would be needed to, when we have web users or other multiprofile nodes.

15:28 <spacekookie> Yea, and another issue is that if you send a message to someone before their announcement has made it to you

15:29 <spacekookie> It's only really something that happens in testing but it should just be journaled until it can be sent

15:29 <spacekookie> Or the user is known

15:29 <spacekookie> I've traced the chat room failure to the alexandria subscription being immediately dropped! That's really weird

15:43 <spacekookie> Okay so I figured out that the subscription is dropped because the chat worker's run condition check immediately returns false, which means it never listens for events

16:24 <mathju> ahh ...

16:25 <mathju> yea, I guess the sending of messages to contacts that have not been announced is really more of a testing then a real world issue :-)

17:08 <spacekookie> I just pushed some code that fixes the immediate issue we had, but some stuff still isn't working. Just to make sure I don't accidentally break the code again

17:51 <spacekookie> So now there's a bit of back and forth message wise before it panics. Getting there :P

19:07 <spacekookie> It works!!

19:32 <spacekookie> mathju: I pushed the working version to the MR branch. Feel free to merge it

19:50 <spacekookie> https://lists.sr.ht/~qaul/community/%3C874ksnk1z3.fsf%40kookie.space%3E

20:02 <mathju> Congrats! Great! Happy to merge it :-)

20:05 <mathju> and thanks for the email!

20:06 <mathju> looking forward to play around and write some tests :-)

20:07 <spacekookie> Hopefully exposing the http api from qauld shouldn't be all that much work, so by the call tomorrow we should have a full stack to play with :)

20:07 <spacekookie> *fingers crossed*

20:07 <mathju> yeay!!! :-)