~singpolyma/cheogram

Only query for disco once

Instead of every time we get told a resource is online.  If we already think it
is online, then assume we already queried and just leave well enough alone.

Cuts CPU usage to normal from insane in production.
Do not report errors that come back from presence discovery
Store presence and caps/disco info in redis
Seperate out mapping for SIP URIs so SMS targets don't get caught
Allow calls from blocked caller id
Hardcode sip.cheogram.com for now

Should this be configurable per-user or just per-instance?
Switch to dhall config and add jingle FT stuff
Switch from many cli args to a dhall config file

Before we add even more config in a sec here
Using newer version of network-protocol-xmpp
Setup for sr.ht CI
New homepage and mailing list workflow
Nuke many log lines

We were always logging effectively in debug mode, but with the amount of active
use we have this was both a possibly privacy issue and make uselessly big log
files.  I haven't looked at most of this in ages, so take it all out until
something calls for it.  Things mostly work these days.
First draft of grouptext porcelein

Works, but does not pass through anything yet if target supports it, and may be
otherwise a bit rough.
Import XMPP internals because we use them

This makes cheogram actually compile with newly released upstream instead of
needing a patched version.
Fixup cabalization
Include node from disco#info query in the reply

As required by https://xmpp.org/extensions/xep-0030.html#info-nodes

Closes #71

Related MR: https://gitlab.com/ossguy/sgx-catapult/-/merge_requests/18
List all needed modules in cabal file
Remove chattiest low-value log
Only say could not join if tried to join
IQ replies should come from the full JID that requests went to
Next