This is a hot issue if anyone's able to give some quick feedback.

Example: http://jsfiddle.net/heff/N8Zs5/
(The spinner will show after the video has ended)

Reported in videojs/video.js#708

My best guess is the cause is the NetStream.Buffer.Empty event handler.

It would make sense that this would fire after the video ends and the buffer is empty, but we shouldn't be firing the waiting event if we're not actually waiting for more data. I don't understand the netstream operation enough to dig much deeper and understand the best way to fix this.

This is a relatively new issue, though it could be caused by the fact that the player no longer manually seeks back to zero when the video ends.

A quick work around in JS layer would be to check the 'ended' property and block waiting events that happen while ended, except that the ended property in HTTPVideoProvider is set to always return false. I'm interested to know if there's a reason for that, because the other providers (Audio/RTMP) will return true when ended.
https://github.com/videojs/video-js-swf/blob/master/src/com/videojs/providers/HTTPVideoProvider.as#L180

Webserver-modules like "ngx_http_mp4_module" or "mod_h264_streaming" allow Flash-clients to seek a video beyond the downloaded buffer line, using a "start"-parameter, e.g.:
http://server.com/path/to/video.mp4?start=10

The article about JWPlayer mentions this feature:
http://www.longtailvideo.com/support/jw-player/jw-player-for-flash-v5/12534/video-delivery-http-pseudo-streaming

Thanks a lot
Markus

other keywords: byte range request

Hi,

in the VideoJSModel.as class, line 280 and 283, i think that the field accessor has a typo.

https://github.com/zencoder/video-js-swf/blob/master/src/com/videojs/VideoJSModel.as#L276

Instead of "volume", "_volume" should be the correct identifier, i think.

Update: That was all wrong. See comments below.

This currently leads to the issue, that sound cannot be turned on again after having it turned off.

There is a related issue in the main videojs project, please set that to fixed as well, once this one is fixed.
videojs/video.js#223

Hello.
Look please whenever possible at adding function of HTTP pseudo-streaming.
I unfortunately don't have knowledge for full implementation of this functionality, but in fact if I am not mistaken - everything simply.

Here an example of implementation from other player:

 if (_pseudoStreamingEnabled) {
      sendEvent(HtmlMediaEvent.SEEKING);
      // Normal seek if it is in buffer and this is the first seek
      if (pos < bufferPosition && _seekOffset == 0) {
        _stream.seek(pos);
      }
      else {
        // Uses server-side pseudo-streaming to seek
        _stream.play(getCurrentUrl(pos));
        _seekOffset = pos;
      }
    }
    else {
      sendEvent(HtmlMediaEvent.SEEKING);
      _stream.seek(pos);
    }

 private function getCurrentUrl(pos:Number):String {
    var url:String = _currentUrl;
    if (_pseudoStreamingEnabled) {
      if (url.indexOf('?') > -1) {
        url = url + '&' + _pseudoStreamingStartQueryParam + '=' + pos;
      }
      else {
        url = url + '?' + _pseudoStreamingStartQueryParam + '=' + pos;
      }
    }
    return url;
  }

Thanks in advance even for attempt to look in this side.

http://help.videojs.com/discussions/problems/1220-black-square-in-flash-file

Play the example video
Either let it finish and replay or scrub around the video.

Expected: No change in playback quality
Actual: Video playback becomes jittery as the video progresses after seek.

In the following example:
https://glomotion.co/video-js.swf?src=https%3A%2F%2Fs3.amazonaws.com%2FGlomotion%2Fmovies%2F3%2Fba0c3086d918f720a09cc1bc592dd43d.mp4&autoplay=true&loop=true

Altough loop is set to true, the video fails to loop.

(Discovered in videojs/video.js#691 and mentioned by bkelley in IRC #videojs)

With the latest control bar updates in video.js core (in master, not released yet) we moved to tracking mouse/touch events to determine when the controls should be visible. Other browsers emit mousemove events from swf element, but Firefox does not, meaning once the controls hide, they never show again.

bkelley has a partial fix but it causes flickering of the control bar (because of mouseouts bubbling from control elements) and this shouldn't work in fullscreen when you can't ever mouse out of the player area.

The real fix will be to bubble mouse move events from the swf to the JS layer. It can probably be implemented in the same way as the stageclick event in the following lines. However we'll need to make sure we're not doubling up the mouse events in browsers that do fire mousemoves for the swf element.

Right click on video while video is playing.

There are two issues related to click events in the swf while a video is being played.

Visual playback stops when the screen is right clicked
But video doesn't actually stop, it keeps rolling.

Right click on video
Select settings
Click in the settings dialog
When adjusting settings, any click on the settings will start/stop video

Hi Mangui , thanks for this wonderful plugin . Does this plugin supports chromium browser ?
I tried this demo http://streambox.fr/HLSProvider/videojs/flash_demo.html in chromium browser but none of the video is played . I installed ffmpeg addon to play mp4 video .

Vulnerable code:
if(loaderInfo.parameters.readyFunction != undefined){
try{
ExternalInterface.call(loaderInfo.parameters.readyFunction, ExternalInterface.objectID);
}
PoC:
video-js.swf?readyFunction=alert%28document.domain%29

Originally reported in videojs/video.js#264

The video starts squashed vertically, but if you got to fullscreen it will fix the aspect ratio.

An example file can be found here:
http://s000.tinyupload.com/index.php?file_id=55943255453027470681

In VideoJS.as, there are these comments:
// we'll want to suppress ANY uncaught debug errors in production (for the sake of ux)
// IEventDispatcher(loaderInfo["uncaughtErrorEvents"]).addEventListener("uncaughtError", onUncaughtError);

In a separate code review, I asked:
Does this suppression prevent the Flash debug dialog from appearing when using the debug version of the player? I'd suggest that we allow errors to surface in the debug player, and if this only applies to using the release player (where I agree, we don't want to surface Flash errors), then ignore this comment.
alemay 4 minutes ago

And Andy answered:
The debug player will still display the typical error modal, unless you call preventDefault() on the error event - which we're doing, and which should probably change.

We should not call preventDefault on error events, so that users can send us as much information as possible when errors (especially uncaught errors) occur.

Found while investigating #30.

I tried creating a hotfix branch to fix that issue, using contribflow, but contrib hotfix create failed with this output:

jwhisenant-retina:video-js-swf jwhisenant$ contrib hotfix start
prompt: Name of the hotfix:  issue-30
-----> Fetching upstream
$ git fetch upstream
From https://github.com/videojs/video-js-swf
 * [new branch]      add-context-menu -> upstream/add-context-menu
 * [new branch]      audio      -> upstream/audio
 * [new branch]      master     -> upstream/master
 * [new branch]      media-source -> upstream/media-source
-----> Creating the stable branch based on remotes/upstream/stable
$ git checkout -b stable remotes/upstream/stable
fatal: git checkout: updating paths is incompatible with switching branches.
Did you intend to checkout 'remotes/upstream/stable' which can not be resolved as commit?
-----> Process aborted
-----> Exit status: 128
-----> You may need to run any following commands manually...
-----> Pushing the stable branch to origin
$ git push origin stable
-----> Tracking the stable branch against origin/stable
$ git branch --set-upstream stable origin/stable
-----> Switching to the stable branch
$ git checkout stable
-----> Updating the stable branch with upstream changes
$ git pull upstream stable
-----> Creating the hotfix/issue-30 branch based on stable
$ git checkout -b hotfix/issue-30 stable
-----> Pushing the hotfix/issue-30 branch to origin
$ git push origin hotfix/issue-30
-----> Tracking the hotfix/issue-30 branch against origin/hotfix/issue-30
$ git branch --set-upstream hotfix/issue-30 origin/hotfix/issue-30
-----> Ready to start building your hotfix in branch hotfix/issue-30.

Open up the player
Go full screen

Expected: smooth playback of video
Actual: very choppy video, audio seems unaffacted

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Brightcove MBR, 16x9, 608k 1096k 1596k 2596kbps renditions, with the lowest rendition giving a 404
Play video

Expected:
Video goes up a rendition to play the video

Actual:
Video does not play any other renditions after starting on the 404

Initially reported in videojs/video.js#513

Flash can sometime take awhile to load a video, and there is no spinner shown in the player to communicate that it's loading. We need to trigger some kind of event that communicates that it's loading/waiting so that the video.js spinner shows.

Hi, is it possible to make it so that it can do fullscreen mode in an iframe in older browsers?
the iframe attribute "allowfullscreen" doesnt work for IE7-9.

So my idea is to have html5 for serious browsers like chrome and firefox, and force flash on IE, because I think flash can run fullscreen inside an iframe without needing to allowing it in the iframe. Youtube does it.

Currently the swf npm package has an install/postinstall script that runs the grunt task to generate the swf from source. The video.js core project will rely on this to get its copy of the swf. This isn't ideal because:

• It make flex-sdk a dependency (not just devDependency), which means it gets download loaded whenever the swf project is installed, and it's 168 MB
• It requires every user to have the grunt CLI already installed
• The task may or may not work on Windows, we need to check

The best approach is probably to do what we've discussed doing with video.js, where you gitignore the dist directory, and git add --force the compiled swf whenever we bump the version of the swf.

Play video and scrub to the end (43-46 seconds)

Expected: video loads at the time scrubbed to.

Actual: video goes back to 42 seconds when scrubbed to any point after.

Note: So long, and thanks for all the fish.

When changing the src of a video, the memory footprint grows linearly.

var interval=2000;
var changeSrc=function(player, src) {
    setInterval(function() {
        player.pause().src(src).play();
    }, interval);
};
videojs("main_video", { "techOrder": ["flash", "html5"] }).ready(function () {
    changeSrc(this, 'http://vjs.zencdn.net/v/oceans.mp4');
});

Example: http://jsfiddle.net/fwcJh/36/ After a minute or so I’m at 2 GB of RAM usage of the Flash plugin.

Hi,

Thanks for adding rtmp support.

what is the format of the rtmp link ? is it documented somewhere ?

thanks

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Select 'Brightcove MBR, 16x9, 608k rendition' test case
Play video
Scrub near to the end of the buffer zone on the video

Expected:
Video plays without any delays

Actual:
Video will wait and load, telling us that the buffer zone is a little inaccurate.

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Brightcove MBR, 21x9, 608k 2596kbps
Throttle bandwidth to 1500kb after the rendition has changed to 2596kbs
Play video

Expected:
Rendition change to 608k

Actual:
No rendition change.

Set video-js-swf and video.js to their feature/hls branches.
Load the player through the server
Click play and open up the developer console
Look for:
[vid1_flash_api] 0 bits per second (bps) [HLS Benchmark]

Expected:
Current bitrate shown

Actual:
A bitrate of 0 shown

http://help.videojs.com/discussions/problems/1361-dynamically-changing-source-in-flash-fallback-fails-in-ie

I am curious to see if there has been any progress with the aspect ratio problem? I have seen multiple threads with no good fix yet, and those were a few months ago. Or is there an estimated time that this will be fixed?

Thanks, Dan.

It would be good to have support for HTTP pseudo-streaming for MP4 and FLV video. Most of HTTP servers today provides this functionality, but videojs player does not handle it.

Hi folks,

I see there is ongoing effort to add HLS support in video.js by rewriting a provider from scratch. I am not clear about the progress of the development, but if by any chance you are interested, I can contribute on providing a HLS provider based on https://github.com/mangui/HLSprovider, which provides an API and a swc file that should ease the work and abstract the mess behind this protocol, supports a bunch of tag, encrypted streams, ...
Cheers,
Mangui

There are a few manifest parsing events that are not logged in the console for the ext-x-stream-inf, the yellow blob of test cases at the bottom of the Sasi - 10/4/13 sheet should cover most of them.

Open the following player (posted at end)
Scroll video to just before the end and let it play to the end

Expected: video player resets video so that it can be replayed with one click
Actual: Player resets video, but is only replayed by clicking twice

(Note: player will show buffer image when scrubbed directly to the end at 46 secs)

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Select any of the following test cases:

broken player in test cases:
renditions 400 error
ts request returns a server timeout
ts request returns a server error
segment request returns a server timeout
segment request returns a server error
highest/lowest rendition return a 404

Expected:
The player should break gracefully (or stay on a working renditon) when it breaks in these cases

Actual:
The player does not break gracefully

Note: Some of these cases need to be played to the buffer zone to see the break, any case that will involve switching to a broken rendition.

I found this while investigating a contribflow bug:
zencoder/contribflow#17

The contrib feature submit call failed because the video-js-swf owner and repo information were supplied as NONE:
user: NONE repo: NONE title: contrib-guide-update body: contrib-guide-update head: BCjwhisenant:feature/contrib-

We should make sure that the owner and repo information for video-js-swf are correct.

If it helps, here's my .git/config information:

[core]
        repositoryformatversion = 0
        filemode = true
        bare = false
        logallrefupdates = true
        ignorecase = true
        precomposeunicode = false
[remote "origin"]
        url = [email protected]:BCjwhisenant/video-js-swf.git
        fetch = +refs/heads/*:refs/remotes/origin/*
[branch "master"]
        remote = origin
        merge = refs/heads/master
[remote "upstream"]
        url = https://github.com/videojs/video-js-swf.git
        fetch = +refs/heads/*:refs/remotes/upstream/*
[branch "feature/contrib-guide-update"]
        remote = origin
        merge = refs/heads/feature/contrib-guide-update

Play a video on the player
Go into full screen
Scrub the player back to the start of the video
Try and get out of fullscreen by hitting the esc key while the video is still playing

Expected: video returns to its default web browser state and continues playing
Actual: video keeps playing in fullscreen

After changing player source from html5 to flash, currentTime passed by SWF always stays zero.

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Brightcove MBR, 16x9, 608k 1096k, 1096k returns a server timeout
Play video to the end of the first 3 TS segments
A rendition switch will then be attempted

Expected:
Error is logged, (playback should either stop or keep playing on current rendition)

Actual:
No timeout error is logged, playback stops, this keeps repeating in the console:
1381256759404 [undefined_component_3_flash_api] HLSVideoProvider: Buffer < 50%!1381256759406 [undefined_component_3_flash_api] ManifestManager.getNextData()1381256759408 [undefined_component_3_flash_api] ManifestManager.getNextData(): SegmentParser.parseSegmentBinaryData() for id: 01381256759409 [undefined_component_3_flash_api] Error #10091381256759410 [undefined_component_3_flash_api] ManifestManager.getNextData() - SegmentParser Error!

And playback eventually resumes.

See videojs/video.js#462

http://jsfiddle.net/xu9Y6/10/ - (view in Firefox where flash player will be used)

I was able to find where the event is triggered, but I can't tell exactly how it's getting triggered.
https://github.com/videojs/video-js-swf/blob/master/src/com/videojs/providers/HTTPVideoProvider.as#L615

My best guess is this is a difference in how Flash works. That it always triggers the metadata event when you start from the beginning of a video file, even if you're just seeking back. Can anyone confirm?

Should the Flash player license be changed to Apache 2.0?

It is forbidden for me to drag the progress forward or backward that i want ,
Using Flash to play MP4 video with VideoJs, why?

What should i do to implement the Flash allow me to choose what startTime to begin?

Just Like Youtube or Vimeo ...

If you change the src of a video the CPU usage grows fast with every src change. You can circumvent this by pausing the video, leading to issue #45.

var interval=2000;
var changeSrc=function(player, src) {
    setInterval(function() {
        player.src(src).play(); // no pause before playing the new one
    }, interval);
};
videojs("main_video", { "techOrder": ["flash", "html5"] }).ready(function () {
    changeSrc(this, 'http://vjs.zencdn.net/v/oceans.mp4');
});

Example (might slow down your PC): http://jsfiddle.net/fwcJh/35/

Set video-js-swf and video.js to their feature/hls branches.
Load the player through the server
Start up Charles and set throttling to 608kbs
Filter Charles to m3u8 and hit play, and then pause when video playback starts
The second m3u8 response will tell you what rendition url was chosen in the overview under URL, look at this and figure out what your current rendition selection is

Expected:
Rendition selection should choose 440kbs

Actual:
Rendition selection chooses 240kbs

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Brightcove MBR, 16x9, 608k 1096k, 1096k manifest request returns a server (error/timeout)
Play video to the end of the first 3 TS segments
A rendition switch will then be attempted

Expected:
Video recognizes the error, logs it to the console and (behavior has yet to have been decided)

Actual:
Video does not log the error, video plays to the the last segment loaded

Note: server error test case does log the error, but the spinner is overlain on the player after the rendition switch attempt.

Play the video.js sample video and let it complete.

Expected: video completes and player is ready to replay video.

Actual: video does not complete, spinner appears over the swf.

We need to add support in the swf for more errors, including:
Network interruption
Unsupported format
Decryption errors (for DRM-protected content)

Set video-js-swf and video.js to their feature/hls branches.
Load the player through the server
Open up the developer console and click play
Look for:
bytes per second (Bps)

Expected:
An accurate current byterate

Actual:
An inaccurate byterate, off by nearly 10x

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Select 'Brightcove MBR, 16x9, 608k rendition' test case
Throttle bandwidth to 600kbs
Play video

Expected:
.... we have no clear expected at the moment

Actual:
The following error shows up in the console when playback starts, and the video has caught up with the buffer zone:

1380897257381 [undefined_component_3_flash_api] ManifestManager.getNextData()index....dlebars1380897257402 [undefined_component_3_flash_api] ManifestManager.getNextData(): SegmentParser.parseSegmentBinaryData() for id: 4index....dlebars1380897257415 [undefined_component_3_flash_api] Error #1009index....dlebars1380897257436 [undefined_component_3_flash_api] ManifestManager.getNextData() - SegmentParser Error!

I'm working on adding audio support to video.js (discussion at videojs/video.js#157) and running into a few speed bumps. One is that an audioloaderror error event is raised when calling vjs_play() while an audio file is already playing. This is done by video.js when seeking.

In HTML5, when you preload a video and no poster is used, the first frame of video shows. This doesn't currently work in the Flash player.
http://jsfiddle.net/heff/S87N2/

Is it possible in Flash?

Start Socrates (make sure it is linked properly to video.js and video-js-swf)
Select any positive path hls test case
Play video
Scrub past the buffer and wait for video to resume playback
When playback resumes, scrub again

Expected:
Video loads and begins playback from the new point scrubbed to

Actual:
Video freezes, and will not play after

Is that possible to rotate the video in the player?

Just to give a background, why it would be handy.
I have a video stream from android phone, which I want to display using video.
According to the documentation,
http://developer.android.com/reference/android/media/MediaRecorder.html

Sets the orientation hint for output video playback. ....
This method will not trigger the source video frame to rotate during video recording,
but to add a composition matrix
containing the rotation angle in the output video if the output format is 
OutputFormat.THREE_GPP or OutputFormat.MPEG_4 so that a video player 
can choose the proper orientation for playback. 

I have more simple task, because I know the orientation of the camera, and I just need to rotate the player accordingly. For HTML5 this is easily done using CSS, and I wonder if this possible to do in a flash player.