smol #realy #NRC #progressreport
https://chat.realy.lol/
did a bunch of mods - popped out the #gio widget/material libraries and tampered with the editor so it does ctrl-del ctrl-backspace to mean "delete to end of line" and "delete to start of line" - also ctrl-left-arrow and ctrl-right-arrow now move the cursor by word (not sure why it wasn't set that way
i can appreciate that at some point this logic should be popped out of the editors so it can be configured by an app, but just fixing some really major key combos i have come to depend on when editing text, as this will be part of the chat text input as well, probably there will be more
anyway, main thing is that those keys work properly now, time to move on to making the actual nav menu contents render properly, again, stealing wholesale from the telegram UI design because i'm a lazy bastard and i like the way telegram's UI works, probably will improve on it later but for now, it's a good model for a decent chat app
oh yeah and i bitched out the gio devs for their totally disrespectful lack of adhering to the concept of stable APIs, i want to say "if you want to break the api, you better be ready to pop a stable version on that branch" but haha keep dreaming, i just need to maintain my widget library's interfacing to the base layout/render pipeline engine for now, it's always fine if it's stable and works everywhere consistently (which it mostly does) they have no excuse for breaking the API anyway, but i'm sure they'll find a reason
but whatever, i can always pin it, so long as it's solid for my purposes i am not going to change the base dependency
properly adding auth to the #realy client library has blatted bugs, both in the no-cgo shitty btcec signer and the nice fast bitcoin core cgo signer
important work for in general, now the spider can *try* to auth to auth-required relays to grab their events, but more importantly, when i move to building the client side of #nrc Nostr Relay Chat (pending nip-79, see pr: https://github.com/mleku/nips/blob/master/79.md https://github.com/nostr-protocol/nips/pull/1654 )
yeah, that has to work
i also will probably need to later fix the relay pool code to work correctly in this way as well
i also had ideas about a one-shot auth scheme where you have a list of relays that you use "your" nsec to sign but if you don't have an account you just make one up and sign with that... this is something i've talked about for some time about the ignorant superstition about auth doxxing users... if i'm paying, they need proof... if i'm not paying, and they demand a signature, i'll just scrawl an x and if that's all they wanted, Good ™️ 🐈
currently testing the spider, first step was making the nip-11 relay information work... blatted several bugs there
then with the relay info, i'm temporarily making the spider only try to use auth-required relays, which led me to the problem of the client auth not being properly implemented yet (i have an entirely different scheme than fiatjaf's versions, which generally involve insecure handling of immutable hex encoded strings, and yeah, right now it's not strongly protected, but at least it's not leaking secrets into garbage memory... i have a plan to add a full memory fence secret key security enclave implementation into the signer, but for now it is at least good enough to not leak secrets around beyond the initialization and config item nsec value in the environment variables
will have this spidering working with auth-required relays soon, it will be interesting to see how many of the auth-required relays will let me query them but at least i'll know that the client code has working auth when they return something other than "gib signature biatch"
Showing page 1 of
1 pages
