200 series and live streaming and llink

[sugatam]

After almost three years, I doubt if Syabas will fix their player to stop using the crazy range requests in the face of a live stream. In fact, why they incur the http request response overhead even in the face of a seekable stream, when they could just be reading the socket, is beyond me, but that is neither here nor there (and FWIW, the 100 series did the right thing in this respect).

Anyway, ludman, I was wondering if you can enhance llink to ignore these range requests when dealing with a live stream (whenever node->cgi_stream is true), and always return the next range of data. To be precise, ignore the offset, but return the amount of bytes requested. That'll open several possibilities, since due to the "external" and "transcode" support, I'm using llink for streaming mlb streams via mlbviewer, to live tv support via myth/llink-mod, etc etc, but only for the 100 series due to the aforementioned bug.

[lundman]

I wrote the "unmms" plug it to work the same way as "unrar" does to MMS streams.

But forcing PCH to stream has not been something I have managed to pull off reliably. Fake-large-size, no Size, Ignore byte-range, Unsupport Byte-range etc. Nothing. They approach Syabas took to the streaming radios is list all the IPs internally. This is rather frustrating. Much much better way is to define a tag, or method, or API, which lets us set streaming mode.

That aside, if you look at the transcoding feature of llink, that spawns the specified script (most often mencoder, or ffmpeg) and sends the output stream to Syabas, llink has the "almost working" logic to make PCH ignore size and seek.

[sugatam]

Thanks for your reply. The transcode mode works perfectly with the 100 series, and since llink is generic in what program it execs to generate the stream, I've been able to use it to call 'mlbviewer' and watch live mlb games on my tv. Again, for the tv that's connected to my A110.

For my other (main) tv that uses the A200, nothing. Looking at the request logs, the A200 is issuing range requests, and llink, rightly, is replying with a 416 Range Unsatisfiable error. There are numerous threads from various people complaining about this over the last 3 years, and clearly Syabas hasn't seen fit to fix this bug, so in my mind the only realistic chance of fixing this is to enhance llink.

If you've played with various approaches and failed with the *200 and live streaming, that's depressing indeed. Is unmms known to work with the 200 series? I'll take a look at that source in that case.

(05-03-2012 07:02 AM)lundman Wrote:  I wrote the "unmms" plug it to work the same way as "unrar" does to MMS streams.

But forcing PCH to stream has not been something I have managed to pull off reliably. Fake-large-size, no Size, Ignore byte-range, Unsupport Byte-range etc. Nothing. They approach Syabas took to the streaming radios is list all the IPs internally. This is rather frustrating. Much much better way is to define a tag, or method, or API, which lets us set streaming mode.

That aside, if you look at the transcoding feature of llink, that spawns the specified script (most often mencoder, or ffmpeg) and sends the output stream to Syabas, llink has the "almost working" logic to make PCH ignore size and seek.

[lundman]

No, I could never get it to work reliably with 200 series. Works with PS3, but that's neither here nor there.

The only fix I can see, is RE on Syabas code and take out the failure jumps it does when it can't seek. Or wait for Syabas to add some way to set streaming-mode