non-splash playlists loop (often omitting last clip)

[phloxic]

As #58 is closed, quoting from #58 (comment)

---

However! When you are on the clip before the last one and search towards the end, it jumps over the last one and restarts where it should switch to the last one.

---

Depending on the last 2 clips, the list might also loop.

Demo: http://flowplayer.blacktrash.org/fp5/playlist.html

As mentioned in #58 playlists work near perfectly in a splash image setup, but now non-splash setups are broken in a different manner.

I will try to make a proposal to parse the playlist items on load into an array and avoid the click() to see whether this improves things.

[phloxic]

Shorter demo: http://flowplayer.blacktrash.org/fp5/poster.html - repeats first clip of two.

[phloxic]

@nnarhinen - on iPad (iOS 6.0.1): when you seek forward within playlist clips, the play button sometimes stays put during the next clip. It also appears during the transition from one clip to another (but that's minor, and probably not fixable).

[phloxic]

I can still reproduce this in Chrome and Safari:
http://example.flowplayer.org/bug/181.html

[nnarhinen]

http://example.flowplayer.org/bug/181.html should be fixed now (using the flowplayer-javascript-playlist.js version of player)

[phloxic]

@nnarhinen - looking good, yipiee ;-)

[phloxic]

@nnarhinen - now http://example.flowplayer.org/bug/181.html hangs in Firefox at end of 1st clip!
Reopening.

[phloxic]

And clicking the html elements has no effect.
@nnarhinen - I still believe that the next and prev methods must be refactored to call load() directly and then take care of the html and css effects instead of trying to short-circuit this with a simulated click:
https://github.com/flowplayer/flowplayer/blob/master/lib/ext/playlist.js#L79
In case of a html playlist, the elements should be parsed and the items stored in array first.
I made several attempts - but lack of time and the fact that the reliance on html is so deeply ingrained thwarted my efforts.
But really, we must get this working reliably. It never did :-(

[nnarhinen]

Is it using the javascript from my attempts to refactor the playlist code?

https://github.com/flowplayer/flowplayer/blob/javascript-playlist/lib/ext/playlist.js#L54As
you can see from there (not merged to dev or master yet) that it
actually does call load now.. (via the playlist added play-method)

2013/2/25 Christian Ebert notifications@github.com

And clicking the html elements has no effect.
@nnarhinen https://github.com/nnarhinen - I still believe that the next
and prev methods must be refactored to call load() directly and then take
care of the html and css effects instead of trying to short-circuit this
with a simulated click:

https://github.com/flowplayer/flowplayer/blob/master/lib/ext/playlist.js#L79
In case of a html playlist, the elements should be parsed and the items
stored in array first.
I made several attempts - but lack of time and the fact that the reliance
on html is so deeply ingrained thwarted my efforts.
But really, we must get this working reliably. It never did :-(

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/181#issuecomment-14044395.

[phloxic]

Is it using the javascript from my attempts to refactor the playlist code?

yes

https://github.com/flowplayer/flowplayer/blob/javascript-playlist/lib/ext/playlist.js#L54As
you can see from there (not merged to dev or master yet) that it
actually does call load now.. (via the playlist added play-method)

Could it be that it just tries to load the mp4 (only one href
fits into an anchor)?

Will try that.

In case of an html setup, it should populate the playlist items
with the types it sees in the video tag - backwards compatibility
etc. ;-)

[nnarhinen]

Oh, so it did that with the "original" way? Ok gonna fix that. Didn't spot
that from the code when refactoring..

2013/2/25 Christian Ebert notifications@github.com

Is it using the javascript from my attempts to refactor the playlist
code?

yes

https://github.com/flowplayer/flowplayer/blob/javascript-playlist/lib/ext/playlist.js#L54As

you can see from there (not merged to dev or master yet) that it
actually does call load now.. (via the playlist added play-method)

Could it be that it just tries to load the mp4 (only one href
fits into an anchor)?

Will try that.

In case of an html setup, it should populate the playlist items
with the types it sees in the video tag - backwards compatibility
etc. ;-)

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/181#issuecomment-14048648.

[phloxic]

I'm not a fan of the html setup ;-) and file name guessing schemes (which won't work with rtmp anyway) - but as we have it ... if you can convince @tipiirai to ditch the current setup, I won't vote against you ;-)

[phloxic]

Will add a JS config to that demo.

[phloxic]

@nnarhinen - done. The JS config works fine indeed.
Again, for compatibility it should probably create an fp-playlist DIV with anchors? respect prev/next anchors? - But it's good to have it working functionality-wise.

[nnarhinen]

Hmm.. yes. It might be wise to create the anchors, maybe disable by some
configuration option? I mean, if we allow creating your own playlist then
it's no idea enforcing our markup.

2013/2/25 Christian Ebert notifications@github.com

@nnarhinen https://github.com/nnarhinen - done. The JS config works
fine indeed.
Again, for compatibility it should probably create an fp-playlist DIV with
anchors? respect prev/next anchors? - But it's good to have it working
functionality-wise.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/181#issuecomment-14049781.

[phloxic]

The Flowplayer Flash playlist plugin in its "internal" deployment method (playlist given in config) does it via expandos.
It's not an immediate requirement - and personally I like to do this myself, customized to the page - but users will ask for the out-of-the-box solution (in spite of potential contstraints). Flowplayer Flash also has the option to configure custom clip properties and that way you can for example pass thumb image src's in the js playlist config. - Endless possibilities, or can of worms, depending how you look at it ;-)

[nnarhinen]

Ok so a1ad78b should fix this (now passes the string directly to load and doesn't try to be too smart). Should be already deployed to example.flowplayer.org for testing

[nnarhinen]

Oh I mean it should fix the stopping after first clip. The anchor creation is still undone, should maybe be moved to #65 (outside the scope of this issue)

[nnarhinen]

Whoops.. loops still (now the second clip..)

[nnarhinen]

Ok now I think I actually got it right.. :)

[nnarhinen]

@blacktrash ping, did you have time to test this?

[phloxic]

@nnarhinen - sorry, I really didn't have any time at all. I hope to get to it over the weekend. I can't see straight anymore, but a quick test of http://example.flowplayer.org/bug/181.html looks very good in Firefox. Functionality-wise it seems it's more or less there, UI-wise: the play button appears briefly during the transition from one playlist item to another, sometimes quite prominently. But don't get me wrong, it's definitely definitely progress.
I've had my share with banging my head on such timing issues by trying to ajax load a source here: http://example.flowplayer.org/scripting/signed-ondemand.html - almost every browser behaves slightly differently, so at least I know to some extent what hard work you put in ;-)

[phloxic]

@nnarhinen - the JavaScript setup loops the playlist; perhaps you haven't handled advance and loop in this case yet?

[phloxic]

@nnarhinen - the prev and next buttons do not work yet either.

[nnarhinen]

They did work earlier and the playlist didn't loop.. live demo?
On 3 Mar 2013 19:49, "Christian Ebert" notifications@github.com wrote:

@nnarhinen https://github.com/nnarhinen - the prev and next buttons do
not work yet either.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/181#issuecomment-14351078
.

[phloxic]

@nnarhinen - the JavaScript setup here: http://example.flowplayer.org/bug/181.html - the HTML setup works now.

[nnarhinen]

Weird.. they were working previously :D gonna look at it

2013/3/3 Christian Ebert notifications@github.com

@nnarhinen https://github.com/nnarhinen - the JavaScript setup here:
http://example.flowplayer.org/bug/181.html - the HTML setup works now.

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/181#issuecomment-14351377
.

[nnarhinen]

Should be fixed now. The playlist wasn't actually in an endless loop, it just looped it once. (It first created the playlist anchors and then looped through them and added them to the playlist..) The second problem actually is just a problem with the playlist css.. the playlist element overflowed the next and prev buttons and prevented clicking on them.. But I now insert the playlist element straight after the video tag to be consistent with the demo.

This should be testable on example.flowplayer.org

[phloxic]

@nnarhinen - confirming the fix, great.
The only nitpick left is the playbutton visibility during item switches in Firefox.

[nnarhinen]

Hmm.. why does firefox behave like that, is it a performance problem or? Is this behavior unacceptable for next release?

[phloxic]

1. users will moan about it, but this should not stop the release ... unless
2. you are able to find a solution for finish event fired more than once #242

My guess is that the play button becomes visible because FF answers to the first finish event - if the problem would belong to the next item we'd see the loading animation, no? Instead we see the a "replay" button for a short interval. Would be nice if you got an answer for #242 (comment)

[nnarhinen]

Unfortunately #242 doesn't seem to be the root to the problem, but it was a good guess. (which actually waits input from @tipiirai , I don't dare remove something I don't know why it has been added in the first place..)

It is the first-frame that seems to show there and for some reason it takes time before the video is playable. :/

[nnarhinen]

#295 created for the remaining issue. Gonna close this because it's not critical anymore and the original issue is resolved.