This commit changes the plugins that uses angular.element to load
resources directly to instead return a function that runs
angular.element. When the browser runs angular.element('<img>') and gets
a src attr it will fetch the content in the src. This makes the
browser running gb automatically fetch all plugin content that gets
linked even if the user didn't request it.
Maybe we could change our plugin architecture to split the match
function and the generator functions so this could be solved more
elegantly.
This is still technically a hack, the elements should really be defined in a view with templates, but it fixes the issue with sanitize where it can't sanitize if it can't parse HTML.
Clicking channel names to join broke because of angular issue 9515.
Reimplement the functionality by binding the scope.openBuffer method to
the window and calling it from an onclick handler for channel name
anchors.
From the comment:
This cache is important for two reasons. One, angular hits it up really often
(because it needs to check for changes and it's not very clever about it).
Two, it prevents weird type conversion issues that otherwise arise in
$store.parseValue (e.g. converting "123." to the number 123 even though it
actually was the beginning of an IP address that the user was in the
process of entering).
Save outgoing /query nicks to a list and then check that list when we
get a buffer title rename and then switch to the buffer if the name
matches any nick in the outgoing list. Reason to use title event instead
of buffer opened event is that buffer open event doesn't contain the
short name. That would require parsing and guessing full names and in
general be more complicated (but maybe more correct)
This patch can be further improved in the future to check for existing
buffers and switch to them if they already exist.
Instead of waiting for the websocket to really close, which can take a
long time because network latency, weechat inresponsiveness etc, we just
set our status to disconnected when user wants to disconnect, and we let
the websocket handle the close in its own time in the background.
If the user wants to reconnect this means there will be a new websocket
connection before the old one has failed, but this works just fine.
