Libflashplayer.so

[sebt3]

^^
This might be the reason :

Getting back on topic:

If anyone else who's had experience doing this kind of black box porting before wants to muck in, I can pm them a link to a tarball of what I've done so far - I've tidied it up a bit & put a getting started script together. It's 516MB though so I don't want to put it out on general access ATM.

I've not yet managed to get any meaningful debugging out of the blasted plugin.

The plugin is recognised by Firefox, and doesn't cause any kind of crash when looking at a flash containing page, so I'm assuming the ABI is ok as is.

I'm currently missing a static initialiser for my shadow __sF, which is unfortunately hidden in the Pandora's C library.

I believe that the plugin is managing to read the swf file, as the pages with flash in them seem to have a correctly sized misbehaving frame - info which I believe comes from the swf.

With the new Firefox plugin wrapper, it's reporting that the child threads terminated, which makes me think that there's something the flash plugin is going to that it's getting an unexpected return result and is therefore dropping out. Which is why I suspect it's an API issue somewhere. I think it's going to be time to hexedi- rename the function calls in the libflashplayer.so file to something random, and put a shim between them and their external equivalents to trace what exactly it thinks it is doing.

Just don't ask me when I'm going to get a round 2it.

emphasis mine

 

[cchiro]

Update:
Thanks to Hitnrun's information, I've been able to run the flash plugin against some controllable code, which has thrown up some very useful debugging. It turns out that the Angstrom folks have been using not only the latest version of the Netscape Plugin API, but also some of their own custom callback values. Wonderful. So the next stage is to import that code in to the test harness framework. Which will eventually have to go in to our version of Firefox. Which probably means ditching Mozilla's livery. What do we call our custom web browser?!?
More as and when.

I have been searching for code to run NSAPI plugins in a test harness (controllable code). Any chance of having a look at your code? I looked thru npplayer.c mentioned above and am still missing a make file and a main() to put it together. Thanks for your interesting posts so far.

 

[Hitnrun]

I have been searching for code to run NSAPI plugins in a test harness (controllable code). Any chance of having a look at your code? I looked thru npplayer.c mentioned above and am still missing a make file and a main() to put it together. Thanks for your interesting posts so far.

You can compile npplayer standalone without a makefile, and it has a main already, just use the compile line I've put on the message.

 

[cchiro]

I have been searching for code to run NSAPI plugins in a test harness (controllable code). Any chance of having a look at your code? I looked thru npplayer.c mentioned above and am still missing a make file and a main() to put it together. Thanks for your interesting posts so far.

You can compile npplayer standalone without a makefile, and it has a main already, just use the compile line I've put on the message.

That sounds great. I am assuming the compile line in the message is using the source from:

svn co svn://anonsvn.kde.org/home/kde/trunk/extragear/multimedia/kmplayer

I haven't had a chance to sync that yet.


By the way, my message was referring to this source:
https://vcs.maemo.org/viewvc/trunk/src/npplayer.c?annotate=826&root=kmplayer

I'd like to have the harness (npplayer) compile and run on MacOS.
Still wondering what hdonk's harness code looks like.
Any thoughts? and thanks.

 

[GizmoTheGreen]

didnt comex make a flash-wrapper thing for iphone/ipad?
that used some other flash module from some other os (android)

i dont recall if it was opensource but if it was, wouldnt it be better than modifying a browser?

edit:
http://github.com/comex/frash

 

[God Ginrai]

didnt comex make a flash-wrapper thing for iphone/ipad?
that used some other flash module from some other os (android)

i dont recall if it was opensource but if it was, wouldnt it be better than modifying a browser?

edit:
http://github.com/comex/frash

Did you not read the topic? That's what they were trying to do, the same as what frash does. There are some problems that are being caused, and they believe that to fix those problems, they need to edit the browser.

-God Ginrai

 

[GizmoTheGreen]

and now we see my overconfidence in alec's stalker page, i totally missed that! D:

have other browsers been tired? :/
sure trying to modify the browser could be a workaround, but the real fix would eventually make it work in any browser

again, sorry *sits in seiza*

 

[God Ginrai]

and now we see my overconfidence in alec's stalker page, i totally missed that! D:

have other browsers been tired? :/
sure trying to modify the browser could be a workaround, but the real fix would eventually make it work in any browser

again, sorry *sits in seiza*

Well, truthfully, the only browsers people are probably going to want to use are FireFox and Chrome, and people seem to be having trouble even getting chrome to compile right, atm. (See the Chrome topic here in dev corner)

-God Ginrai

 

[Tempel]

and now we see my overconfidence in alec's stalker page, i totally missed that! D:

have other browsers been tired? :/
sure trying to modify the browser could be a workaround, but the real fix would eventually make it work in any browser

again, sorry *sits in seiza*

Well, truthfully, the only browsers people are probably going to want to use are FireFox and Chrome, and people seem to be having trouble even getting chrome to compile right, atm. (See the Chrome topic here in dev corner)

-God Ginrai

No man. I use Midori everywhere. Like from my desktop, right now. I'll end up using it on the Pandora too, just because it's easier to sync my bookmarks that way.


Gizmo, if you look back a bit in the thread, you'd see:

It turns out that the Angstrom folks have been using not only the latest version of the Netscape Plugin API, but also some of their own custom callback values. Wonderful. So the next stage is to import that code in to the test harness framework. Which will eventually have to go in to our version of Firefox.

So apparently the issue is inherent in Angstrom, not the methods used to wrap the plugin. But a proper permanent solution would be to push a proper NPAPI to Angstrom. Or so I've gathered, I'm not exactly fluent on what's transpiring here.

 

[b1ueskycomp1ex]

Ok, my recommendations for names a pandora browser could use:

Pandora's Box
Panzilla (lol...)
FireBox




Ok, I'm out of ideas. >>

But I rather like Pandora's Box ;D

 

[God Ginrai]

and now we see my overconfidence in alec's stalker page, i totally missed that! D:

have other browsers been tired? :/
sure trying to modify the browser could be a workaround, but the real fix would eventually make it work in any browser

again, sorry *sits in seiza*

Well, truthfully, the only browsers people are probably going to want to use are FireFox and Chrome, and people seem to be having trouble even getting chrome to compile right, atm. (See the Chrome topic here in dev corner)

-God Ginrai

No man. I use Midori everywhere. Like from my desktop, right now. I'll end up using it on the Pandora too, just because it's easier to sync my bookmarks that way.

Except for the fact that the Pandora Midori crashes too often to be reliable. That's why they changed the default browser to Arora. It was stable. If you persist to use Midori, that's your choice, but I'm pretty sure most people will go for what is stable.

-God Ginrai

 

[Tempel]

Except for the fact that the Pandora Midori crashes too often to be reliable. That's why they changed the default browser to Arora. It was stable. If you persist to use Midori, that's your choice, but I'm pretty sure most people will go for what is stable.

I've been using Midori long enough to be well aware of it's stability issues (though it's rapidly improved, and the Pandora version should see a significant update once someone merges new Angstrom updates into Pandora sources; I'm looking into it myself).

My point was simply that there are lots of browsers; people aren't only going to want to use Firefox or Chromium. Being able to use Flash, even if only in modified Firefox and Chromium, will be great, but ultimately getting it running in any unmodified browser will be best. There are lots of browsers available (eg. Arora that you mentioned), so sticking with established standards will allow the flexibility for users to pick any browser they want.

But this discussion doesn't really matter anyways. Devs, you work your magic; once Flash gets running, even in a limited fashion, we can then worry about what else is possible.

 

[hdonk]

I've been using frash for part of the wrapper. Unfortunately just had a major crunch on at work and had to put it on the back-burner. I should be able to spend some more time on it in the next couple of days.
I does look as if it should be possible to have enough of a wrapper to make it browser agnostic, but until I know all the things that need changing, I won't know for sure.
It's a shame frash is written for the iPad OS - that's what's making it impossible to use straight in my port. However as Android is more closely related to Linux mostly it's pruning code and importing stuff from Android.
The good news is with a pause, frash has been improved - but whether those improvements help me or not, I don't yet know!

 

[joseluisjazz]

Except for the fact that the Pandora Midori crashes too often to be reliable. That's why they changed the default browser to Arora. It was stable. If you persist to use Midori, that's your choice, but I'm pretty sure most people will go for what is stable.

Funny. Arora keeps crashing on me. And it doesn't even remember the tabs I had when it crashed.

 

[God Ginrai]

And it doesn't even remember the tabs I had when it crashed.

Most browsers didn't even think to start doing that until just recently. I'm not surprised Arora doesn't do that.

-God Ginrai

 

[sindbad]

I don't know if anyone else tried, but wouldn't Maemo's flash be a better fit, since it already works in a xulrunner browser?

 

[God Ginrai]

I don't know if anyone else tried, but wouldn't Maemo's flash be a better fit, since it already works in a xulrunner browser?

Someone tried. I believe the topic was in GenTalk. The end result is that they kept running into dependencies for Maemo-specific things.

-God Ginrai