@virtuacoplenny, the changes of this commit are not necessarily in
source code that you introduced in
https://github.com/jitsi/jitsi-meet/pull/1499 but I saw violations in
files modified in the PR which I had to read in order to understand the
PR.
Converting the invite modal includes the following:
- Creating new react components to display InviteDialog. The
main parent components are ShareLink and PasswordOverview,
the later handles displaying lock state and password editing.
These components do not make use of atlaskit as the component
for input does not yet support readonly, so for consistency
within the modal content no atlaskit was used.
- Using redux for keeping and accessing lock state instead of
RoomLocker.
- Publicly exposing the redux action lockStateChanged for direct
calling on lock events experienced on the web client.
- Removing Invite, InviteDialogView, and RoomLocker and references
to them.
- Handling errors that occur when setting a password to preserve
existing web funtionality.
When a dialog is opened on Android, full-screen mode is exited but we (the app)
know nothing about this. Make sure we re-enter full-screen mode once a dialog is
closed, if the conditions to be in such mode are still met.
Instead of using AtlasKit Single-Select, use Dropdown. Dropdown
differs in that an icon can be specified for the trigger element,
whereas Single-Select currently supports icons for all elements,
and Dropdown can show all options incuding the already-selected
option.
This change does introduce the issue of the trigger element not
taking up 100% width of the parent. Supporting such would involve
overriding AtlasKit CSS. The compromise made here was to do a
generic override of max-width so the trigger elements at least
stay within the parent and aligning the trigger elements to the
right.
Some atlaskit components, such as field-text, inherit text color.
This is a problem with components that are white as they will
inherit $defaultColor, which is a light gray. So instead, for
the atlaskit modal, set a color for all the form content so it
can be inherited instead.
Dialog does not currently support displaying dynamic strings
for titles, only static strings listed for translation. Accept
a new prop that explicitly states it is for setting the title
and have the web dialog prefer it over the titleKey.
When the prosody setting has muc_allowners, everyone joins as a
moderator. In this case, the local user will not be set as a
moderator in the redux store as the USER_ROLE_CHANGE event will
trigger with the local user id before the redux store has set
the actual local user id--something that happens on
CONFERENCE_JOINED. The fix is to explicitly signal the local user
role has changed to the redux store, which follows the
implementation of pre-existing web logic.
Device Selection re-uses the local video track for its preview.
When displaying Device Selection while video muted, the text
"muted" displays within the video preview, but some translations
may mistake this to mean audio muted. The text has been changed
to be explicit about video mute. This is a temporary solution;
at some point Device Selection should not re-use the local track
except for IE, the one browser that cannot get multiple tracks
from the same video input.
Over time features/base/util became a bucket where people seemed to dump
just about anything they couldn't think of a better place for. That's a
trend I don't like encouraging. Given that roomnameGenerator.js is
currently used in features/welcome only, I'm fine with moving it there
for the greater good.
Because timeUtil.js computes hours, minutes, and seconds out of a single
time/duration using three separate functions, I wouldn't recommend using
it, especially reusing it. That's why I'm even making the functions
private to their current use location.
I don't like the file/function name, I'm not excited about the
complexity of the logic it implements, and it's definitely a reusable
piece worthy of being called a utility.
Move logic to open device selection outside of SettingsMenu so
it can be called independently by either SettingsMenu or by
the settings button itself if no other settings but devices will
be displayed.
Saúl Ibarra Corretgé reported that Recording shows an error dialog
stating "There was an error connecting to your camera". Hristo Terezov
and Yana Stamcheva traced that the problem originates in
da4425b5c0
and, more specifically, is caused by a different order of execution due
to the move of the invocation of the function Recording.init.
The solution is to bring back the execution location of Recording.init.
Atlaskit at times will have localized styling for font-size and
sometimes will not. The button component will inherit its
font-size whereas selectors have localized font-size of 14px. For
consistency, the cancel/submit buttons on the atlaskit modals
will also have 14px. The atlaskit story book examples also use
buttons with 14px font-size.
For images < 80 of size forder radius doesn't work properly (it looks like a
square with rounded corders), however, using a duble sized radius does the
trick. Go figure.
The Device Selection modal consists of:
- DeviceSelection, an overly smart component responsible for
triggering stream creation and cleanup.
- DeviceSelector for selector elements.
- VideoInputPreview for displaying a video preview.
- AudioInputPreview for displaying a volume meter.
- AudioOutputPreview for a test sound output link.
Store changes include is primarily storing the list of
available devices in redux. Other app state has been left
alone for future refactoring.
We seemed to be using the names "film strip" and "filmstrip" (and,
consequently, their source code-conscious forms such as film-strip,
FilmStrip, etc.) In order to comply with our coding style which requires
a consistent one name for a given abstraction, choose one name and
rename the uses of the other name.
Wikipedia has a definition of a "filmstrip", I couldn't find a "film
strip". I guess our abstraction can be seen as what's described there.
When I google "film strip", I get results about "filmstrip" at the top.
That's why I chose "filmstrip".
Certain uses of "film strip" such as interfaceConfig.filmStripOnly and
in the external API I left untouched in an attempt to preserve
compatibility.
I wasn't sure whether CSS was tangled in compatibility so I made a
choice and renamed there was well.
When a conference is to happen in a domain which is not the defaut, its config
is loaded and set. As part of this process, lib-jitsi-meet is disposed. Because
disposing is asynchronous, events happen in this sequence:
- set new config
- dispose lib (which effectively wipes the config)
- init lib
This results in the library to be initialized without the loaded config, which
was lost. This commit fixes that by delaying setting the config and
re-initializing the library until it was disposed.
The functionality to use the react-native-webrtc custom API for fast switching
cameras was moved to JitsiLocalTrack in lib-jitsi-meet. Use that.
Ref: https://github.com/jitsi/lib-jitsi-meet/pull/444
The behavior can be triggered with the toggleAudioOnly action, which is
currently fired with a button.
The following aspects of the conference will change when in audio only mode:
- local video is muted
- last N is set to 0 (effectively muting remote video)
- full-screen mode is exited
- audio mode is set to "audio chat" (default output is the earpiece)
- the wake lock is disengaged
One aspect not handled in this patch is disabling the video mute button while in
audio only mode. The user should not be able to turn back video on in that case.
They better represent if a participant has video available or not. There are
cases when even a participant in the last N set would not have video because it
disconnected momentarily, for example.
Use the curstom _switchCamera API provided by react-native-webrtc to toggle the
camera instead of destroying the current track and creating a new one.
_switchCamera is implemented at a low level, so the track perceives no changes,
thus being a lot faster and less involved since the capturer doesn't need to be
destroyed and re-created.
In addition, don't mirror the video for the back camera.
Ref: https://github.com/oney/react-native-webrtc/pull/235
When a new local video track is created an associated video capturer is created
for it. The cause for the freezes seems to be creating mutliple tracks (which
come with a video capturer each). Fix this by first disposing of the previous
video track before creating the new one.
Ref:
https://github.com/oney/react-native-webrtc/issues/209#issuecomment-281482869
Lib-jitsi-meet does not really implement isScreenSharing. Besides,
getCameraFacingMode will already make sure that the video track does not
represent a desktop stream.
It got broken while rewriting the Web toolbar in React Toolbox. There is
a problem with the toolbars and how we construct the intialState of the
buttons. The _getInitialState() in the toolbox reducer gets the list of
buttons from interfaceConfig, but in fact interfaceConfig is meant to be
overriden in several very important cases. One of the cases being the
external API, which we use in several projects in production.
JSDoc comments didn't follow the ESLint rule for properly formatted
sentences.
BTW, I'm not blind to the fact that PasswordRequiredPrompt and
RoomLockPrompt participated in a birthing of source code through
copy+paste. (If we do not copy+paste, we will not have to fix one and
the same source code such as comments multiple times.)
Lyubo Marinov
Leonard Kim
Saúl Ibarra Corretgé
yanas
Ilya Daynatovich