Update storing of app settings. Store settings in a closured
`app_settings` object inside `@converse/headless/shared/settings`
Remove the `_converse.settings` object.
The eventual goal is to avoid UI-related stanza processing if the relevant chats
aren't in the DOM.
With the current architecture, chatboxes are created (and the stanzas
related to them processed) even if `#conversejs` isn't in the DOM.
* Initial work on making controlbox an element
* Create a shared base class
* Ceate ChatBoxViews proxy
* Update sass now that certain classes are moved to converse-chats element
In some cases a race condition can occur where a new message is received
before the MAM query starts.
Previously, the newly received message would be considered the latest
message to query from, thereby causing a gap in the history.
- Add test for incoming RAI message
- Only enable RAI if the user is affilated in MUC being left
- Handle error presence indicating a resouce-constraint
- Don't unregister stanza handlers in `leave`, since we still want to
listen to RAI-related stanzas. Instead unregister upon the `destroy`
event.
- The `converse.api.query` method now no longer accepts an RSM instance.
- The RSM class now separates `query` parameters from `result` attributes
- Improve JSDoc docs and remove need to make `converse-rsm` a plugin
- Add typedefs for the options expected by RSM and `api.archive.query`
If a message comes in after joining the MUC but before the MAM messages
have been fetched, then converse-mam will query for MAM messages newer
than that message, causing an empty history (except for that one message).
by checking for the affiliation data on the probe response presence and
not on the occupant model (which might be created from a message and
therefore not yet have up to date affiliation data).
It's better to parse an incoming message stanza early, than to have
all kinds of methods throughout the codebase that does querySelector
etc.
Firstly, it allows us to catch and report errors and malicious stanzas early on.
It also simplifies programming because you don't need to try and
remember how to properly parse a stanza, all the work is done upfront
for you.
