Up until now we relied on implicit loading of middlewares and reducers, through
having imports in each feature's index.js.
This leads to many complex import cycles which result in (sometimes) hard to fix
bugs in addition to (often) breaking mobile because a web-only feature gets
imported on mobile too, thanks to the implicit loading.
This PR changes that to make the process explicit. Both middlewares and reducers
are imported in a single place, the app entrypoint. They have been divided into
3 categories: any, web and native, which represent each of the platforms
respectively.
Ideally no feature should have an index.js exporting actions, action types and
components, but that's a larger ordeal, so this is just the first step in
getting there. In order to both set example and avoid large cycles the app
feature has been refactored to not have an idex.js itself.
* ref: disable ICE restart by default
The reason for that it's currently causing issues with signaling when
Octo is enabled. Also when we do an "ICE restart"(which is not a real
ICE restart), the client maintains the TCC sequence number counter, but
the bridge resets it. The bridge sends media packets with TCC sequence
numbers starting from 0.
The 'enableIceRestart' config option can be used to force it, but it's
not recommended.
Move all polyfills to a standalone feature, which gets imported before anything
else in the mobile entrypoint. This guarantees that any further import sees the
polyfilled environment.
In
1ffd75c0a6
we switched to using the localStorage wrapper provided by js-utils, which
checks for window.localStorage's availability very early. Our polyfill must be
applied earlier that any such import.
Here we are importing it in the entrypoint, which means no code ran before this,
literally.
We are downloading code off the Internet and executing it on the user's device,
so run it sandboxed to avoid potential bad actors.
Since it's impossible to eval() safely in JS and React Native doesn't offer
something akin to Node's vm module, here we are rolling our own.
On Android it uses the Duktape JavaScript engine and on iOS the builtin
JavaScriptCore engine. The extra JS engine is *only* used for evaluating the
downloaded code and returning a JSON string which is then passed back to RN.
React Native doesn't define __filename nor __dirname so do it artisanally. In
addition, this helps with centralizing the configuration passed to loggers.
Using anything non-serializable for action types is discouraged:
https://redux.js.org/faq/actions#actions
In fact, this is the Flow definition for dispatching actions:
declare export type DispatchAPI<A> = (action: A) => A;
declare export type Dispatch<A: { type: $Subtype<string> }> = DispatchAPI<A>;
Note how the `type` field is defined as a subtype of string, which Symbol isn’t.
Instead of bundling it in lib-jitsi-meet, which unnecessarily increases
lib-jitsi-meet's bundle size, polyfill it here so it's available in the global
scope, just like the web does.
* feat: Displays the E2E RTT in the connection stats table.
* fix: Whitelists the ping config properties.
* ref: Addresses feedback.
* npm: Updates lib-jitsi-meet to e097a1189ed99838605d90b959e129155bc0e50a.
* ref: Moves the e2ertt and region to the existing stats object.