Complete failure in Opera with Flash 11,4,402,265

[GoogleCodeExporter]

Quote http://flowplayer.org/forum/8/102899:

"""
Chances are high nobody reports it here, but I just downloaded Flash 
11,4,402,265. Works the same in all browsers except for Opera. All Flowplayer 
demo now fail in Opera (tried 11 and 12). JWPlayer runs fine.

Tried to update Flash in Ubuntu. Will not work: "Flash Player 11.2 is the last 
supported Flash Player version for Linux. Adobe will continue provide security 
updates." Flowplayer still plays in Opera on Ubuntu.
"""

Original issue reported on code.google.com by blacktrashproduct on 23 Aug 2012 at 12:21

[GoogleCodeExporter]

Some of their policies make me wish Macromedia didn't get bought out :o here is 
more information about that one

http://www.omgubuntu.co.uk/2012/02/adobe-adandons-flash-on-linux

That seems to amaze me google who is trying to also make flash redundant is now 
providing it as an alternative plugin for chrome on linux :)

Also what OS is opera running on. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 4:30

[GoogleCodeExporter]

I first discovered this bug on Windows 7. Later I also upgraded Flash on XP 
Pro. Same failure.

I have got OSX Snow Leopard running in a Virtualbox, but that never went 
smoothly. Always sound only and white screen. After upgrading I still get the 
sound in JW Player, but not in Flowplayer, so there is a change the same bug is 
present in OSX as well.

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 7:35

[GoogleCodeExporter]

Confirming there is an issue with the latest update of flash with latest opera 
on windows 7. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 7:45

[GoogleCodeExporter]

It looks like a bug in the browser. If i load the file directly, it requires to 
refreshes to get the loader initialised and display the player. I have to load 
the file directly 2 times before being able to make it load in the browser 
correctly. If i clear the cache same problem again. Needs some kind of work 
around in the loader I suspect. This browser is very hard to support and its 
debugging tools is buggy itself it seems. Sadly it has it's tentacles into 
portable devices. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 8:01

[GoogleCodeExporter]

I ran another instance of OSX Snow Leopard in a virtualbox. Before the update 
to Flash 11.4: no problems in Opera. After the flash update: everything normal 
in JW Player, black screen in Opera. 

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 8:18

[GoogleCodeExporter]

Have you made sure that jwplayer's flash mode is enforced? Only a flv or mp4 
file configured? Just to make sure it's not going into html5 mode.

Original comment by blacktrashproduct on 23 Aug 2012 at 9:04

[GoogleCodeExporter]

Yes. I made sure. Clicked right, got the Flash settings and 'About Adobe Flash 
Player 11.4.402.165...'

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 9:23

[GoogleCodeExporter]

I can't pin point the exact source of the bug yet but the loader is not 
reporting the bytes total of the player therefore it will not reach 
initialisation. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 9:37

[GoogleCodeExporter]

Same deal with osx opera 11.4. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 9:45

[GoogleCodeExporter]

looks like how its embedded into the browser. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 10:02

[GoogleCodeExporter]

Confirming this is a serious bug with Opera and not a Flowplayer issue. I am 
unable to get bytestotal information from normal flash files. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 10:38

[GoogleCodeExporter]

jwplayer does not use a preloader at all, it doesn't wait until the main 
application has loaded to begin intialization hence why its working. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 10:42

[GoogleCodeExporter]

Player loads fine avoiding the loaded bytes check, not sure what to advise I 
think the loader check is a good idea to keep. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 10:45

[GoogleCodeExporter]

Providing a workaround by also checking for a complete event works fine. The 
other solution is to fix IE bugs, this solution to fix Opera bugs :)

Original comment by electrot...@gmail.com on 23 Aug 2012 at 10:58

[GoogleCodeExporter]

OK that works fine. 

https://dl.dropbox.com/u/3394987/flowplayer.commercial-3.2.14.zip

Please try this. Might need a bit of testing as there is now two different load 
completion checks. The original one to fix the IE issues, and the new one to 
fix the Opera issues which will turn off the first one if this comes first. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 11:06

[GoogleCodeExporter]

This may fix the gzip issues also, I believe thats why it hasn't been working 
because the bytestotal is zero. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 11:07

[GoogleCodeExporter]

Works in my Opera in Win7 (except for the known sound issue):
https://sessiondatabase.net/flowplayer/start7m_comm.php

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 11:29

[GoogleCodeExporter]

It seems a bit weird that Opera would be at fault here given that it worked 
with previous version of Flash. I will investigate it from the browser side. 
Can you point me to some demo/TC that uses flowplayer and doesn't have a 
workaround yet?

BTW. I work for Opera.

Original comment by rch...@gmail.com on 23 Aug 2012 at 11:35

[GoogleCodeExporter]

https://sessiondatabase.net/flowplayer/start7m.php
does not play in Opera, but also has the sound issue.

See http://flowplayer.org/forum/8/102841 for the sound issue (in most browsers)

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 11:56

[GoogleCodeExporter]

Ahh @ rchl2k I have absolutely no idea. If I launch opera via a simple flash 
builder app, the loaderInfo.bytesTotal is always zero so the same. Because of 
this the player will not initialise, the bytesLoaded is filled up but because 
bytesTotal is always zero it will not continue. The work around is to listen 
for the complete event also. The reason being the original code checks for the 
loaded bytes on an enterframe event because of a bug in IE not dispatching 
complete events.

Original comment by electrot...@gmail.com on 23 Aug 2012 at 12:09

[GoogleCodeExporter]

I tried to match the embed object the same as jwplayer directly outside the js 
api so it's not an embedding issue or perhaps it could be still. If you load 
flowplayer directly , refresh a few times it will load properly the first time 
with a cleared cache it will not because of this issue, but my fix should work 
around it, OR if you can track it down then there shouldn't be a need for the 
fix. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 12:11

[GoogleCodeExporter]

the zero bytesTotal thing btw is also caused by gzip headers thats why people 
need to turn it off for swf files but this isn't the case because im loading it 
locally. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 12:12

[GoogleCodeExporter]

@Jan have you tried the updated file in the zip at all ? 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 12:26

[GoogleCodeExporter]

Yes. See comment 17.

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 12:32

[GoogleCodeExporter]

Thanks confirming that works fine. Can you try and take out the start 
parameter, and duration properties. I don't get sound on chrome either. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 12:42

[GoogleCodeExporter]

Please do not mix this with issue538 which concerns Wowza + clip.start + Flash 
> 11.1.

Original comment by blacktrashproduct on 23 Aug 2012 at 12:47

[GoogleCodeExporter]

Without the sound issue is this one:
https://sessiondatabase.net/flowplayer/start_comm.php

Original comment by Jan.Ehrhardt@monitor.nl on 23 Aug 2012 at 1:00

[GoogleCodeExporter]

Again it's fine ;) @ rchl2k would you be able to confirm a few things before I 
make adjustments on our end to resolve the issue. 

Original comment by electrot...@gmail.com on 23 Aug 2012 at 1:46

[GoogleCodeExporter]

No reaction from rchl2k yet !? Are you planning a new release with the fixes?

Original comment by Jan.Ehrhardt@monitor.nl on 24 Aug 2012 at 10:07

[GoogleCodeExporter]

Once tested i think its safe to go with both preloader methods. 

Original comment by electrot...@gmail.com on 24 Aug 2012 at 11:28

[GoogleCodeExporter]

Can I be of help with testing it?

Original comment by Jan.Ehrhardt@monitor.nl on 24 Aug 2012 at 12:28

[GoogleCodeExporter]

https://dl.dropbox.com/u/3394987/flowplayer.commercial-3.2.14.zip Yes this is 
the file with the fix. 

Original comment by electrot...@gmail.com on 24 Aug 2012 at 12:39

[GoogleCodeExporter]

Yes, that is the one I am already using in
https://sessiondatabase.net/flowplayer/start_comm.php

It is just working. In IE and in Opera.

Original comment by Jan.Ehrhardt@monitor.nl on 24 Aug 2012 at 1:18

[GoogleCodeExporter]

OK so it loads in Opera now is that correct ? 

Original comment by electrot...@gmail.com on 24 Aug 2012 at 2:07

[GoogleCodeExporter]

Yes. In every Opera I tried. 11 & 12 on Windows, 12 on Ubuntu, 12 on MAC OSX 
Snow Leopard. And even in Opera Mobile on Android ICS.

Original comment by Jan.Ehrhardt@monitor.nl on 24 Aug 2012 at 2:18

[GoogleCodeExporter]

woah thankyou 

Original comment by electrot...@gmail.com on 24 Aug 2012 at 2:40

[GoogleCodeExporter]

Sorry, I've failed to star the issue and did not get any reply notifications. I 
will still have a look to see why it stopped working in Opera with new Flash 
version.

Original comment by rch...@gmail.com on 25 Aug 2012 at 3:42

[GoogleCodeExporter]

its reporting 0 for bytesTotal pretty much which is apparantly the same problem 
if swf is sent gzipped. Any flash app will report that. the work around is to 
check for a complete event which apparantly is broken in IE so it's needs both 
checks ..

Original comment by electrot...@gmail.com on 25 Aug 2012 at 5:17

[GoogleCodeExporter]

Opera is reporting correct stream size in both cases (Flash 11.3 & 11.4). The 
bug that Opera has is that it does not report original server headers when 
resource is cached. Apparently Flash 11.3 did not care about that (or had 
workaround) while 11.4 does care.
That bug is already fixed in Opera 12.50 (not a final version yet). We'll try 
to get it out for 12.02 too which will be released very soon.

Original comment by rch...@gmail.com on 25 Aug 2012 at 10:44

[GoogleCodeExporter]

Hi thanks this is for local resources too though, perhaps it's the same issue. 
I found though I had to get it cached for it to produce the bytesTotal, strange 
it's the other way ? 

Original comment by electrot...@gmail.com on 26 Aug 2012 at 3:14

[GoogleCodeExporter]

@rchl2k: Fixing it only in 12.02 and up forces me to choose between two bugs: 
the Flash bug in 11.61 and this one:
http://my.opera.com/community/forums/topic.dml?id=1353052&t=1345975859&page=1#co
mment12163532

Even on my Ubuntu and MacOSX Virtualboxes I quickly went back to 11.60/61 to 
avoid the 'single file bookmark' bug. Rijk once commented he could not 
reproduce it, but it is definitely there.

@the flowplayer devs: do not fix it for Opera 12.02+ only. Youc current fix (in 
the dropbox) works for Opera 8, 9, 10, 11 and 12.02 as well. I did not try 
12.50 yet, but will look into that.

Original comment by Jan.Ehrhardt@monitor.nl on 26 Aug 2012 at 10:22

[GoogleCodeExporter]

I can confirm that the Flash bug was fixed in Opera 12.50, tested on this one:
https://sessiondatabase.net/flowplayer/ (which does not play in Opera 11.61).

But the single folder bookmark bug is still present in Opera 12.50: I cannot 
browse to Projecten > FFmpeg > Hardware for instance. Ergo: I still cannot 
upgrade my Opera 11.61.

Original comment by Jan.Ehrhardt@monitor.nl on 26 Aug 2012 at 10:37

[GoogleCodeExporter]

@electroteque, I've checked that case too now. One thing worth to mention is 
that local files (file:// protocol) are never cached in Opera. But obviously 
local files also don't send any headers. And here appears to be a slight 
browser compatibility problem because we (Opera) still send one "fake" header 
(HTTP/1.1) while Chrome (for example) sends NULL. I'm talking about it from the 
NPAPI point of view (see NPStream->headers in [1]), not how it's interpreted 
and exposed to authored SWF application by Flash.

@Jan, I believe there is no doubt that flowplayer should workaround this issue 
so that it works in all version. I'm talking about a fix from the Opera's point 
of view which we should do regardless.

[1] https://developer.mozilla.org/en-US/docs/NPStream

Original comment by rch...@gmail.com on 26 Aug 2012 at 10:42

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

What do you suggest, we wait or go ahead with the work around ? 

Original comment by electrot...@gmail.com on 27 Aug 2012 at 4:31

[GoogleCodeExporter]

Good news everyone, just autoupdated Opera 12.01 x64 to 12.02, and Flowplayer 
is working again with Flash 11.4.402.265 (tested with Flowplayer 3.2.5 and 
3.2.14 builds).

Original comment by timo.va...@gmail.com on 30 Aug 2012 at 8:27

[GoogleCodeExporter]

Ok so the fix is not neccesary anymore i will try. 

Original comment by electrot...@gmail.com on 30 Aug 2012 at 8:35

[GoogleCodeExporter]

No, the fix is still needed. See comment 41, 42 and 43.

Original comment by Jan.Ehrhardt@monitor.nl on 30 Aug 2012 at 9:09

[GoogleCodeExporter]

The Opera update addressing the problem was released but it obviously doesn't 
mean that everyone will get updated version immediately. So the workaround on 
the flowplayer side would still be appreciated.

Original comment by rch...@gmail.com on 30 Aug 2012 at 9:13