Roku Developer Program

[jamesonf]

@RokuNB,

The issue was never resolved. We have a few Rokus on the beta receiving the beta firmware updates as well.  We have been working hard at getting an RSG build running stable, and we just submitted it today to Roku.

As far as we can tell, Scene Graph hasn't crashed using the same api endpoints, videos, and codecs.

[JordanT92]

I'm having the same issue on a few boxes running software I'm developing. Happens at random and can be stopped by quickly pressing the back button closing the video node. If you leave it a few seconds it no longer responds and has to be restarted.

The issue has manifested on multiple Roku 2's, multiple Roku Streaming Sticks (3500EU) but not on any Roku Express boxes (imported to UK from US).

Only started happening in the last few weeks, coinciding with the latest update.

[destruk]

For Roku3 you might want to test firmware 7.7.0 build 4117-04

I don't have a working roku2 anymore to test with.

[NB_]

"jamesonf" wrote:
The issue was never resolved. We have a few Rokus on the beta receiving the beta firmware updates as well.  We have been working hard at getting an RSG build running stable, and we just submitted it today to Roku.

As far as we can tell, Scene Graph hasn't crashed using the same api endpoints, videos, and codecs.

Yes, i noticed you have been pushing RSG build but it might not be the difference between SDK1 and RSG. In your initial report (8/4) you pointed it's happening with your RSG code too. I don't know of any change made here that would have resolved _only_ the RSG side.

But was a specific change we made on our side on Monday that may have fixed both cases. Can you very specifically ask around and see if the issue has happened since Tuesday in your production SDK1 app?

[JordanT92]

"RokuNB" wrote:

"jamesonf" wrote:
The issue was never resolved. We have a few Rokus on the beta receiving the beta firmware updates as well.  We have been working hard at getting an RSG build running stable, and we just submitted it today to Roku.

As far as we can tell, Scene Graph hasn't crashed using the same api endpoints, videos, and codecs.

Yes, i noticed you have been pushing RSG build but it might not be the difference between SDK1 and RSG. In your initial report (8/4) you pointed it's happening with your RSG code too. I don't know of any change made here that would have resolved _only_ the RSG side.

But was a specific change we made on our side on Monday that may have fixed both cases. Can you very specifically ask around and see if the issue has happened since Tuesday in your production SDK1 app?

If it helps, I've had the issue happen yesterday on a Roku 2 running the latest firmware, it happened about 20 minutes into a test MP4 I let run (h264 aac). This was with a RSG video node. I pressed back after a few seconds and it went back to the previous screen without issue, from there I was able to play the video again. Usually it would crash the entire box though and it would have to be restarted.

[NB_]

Those still experiencing the issue -
have you noticed some relation between the size of the video file played - or position in the video file - and the issue happening?
E.g. "happens only on files >2gb", or >4gb etc. Or does it happen (more) reliably when fast-forwarding past particular position in a big file and then playing - even if did not play that long time?

[NB_]

"jamesonf" wrote:

When I sideloaded the channel I was able to view the console as it crashed. The video component reported an unknown event, but it was still reporting to our Google Analytics. By “unknown” event - it was not a documented event within here: https://sdkdocs.roku.com/display/sdkdoc/roVideoPlayerEvent[/url:5z0m5pwr]

Follow-up on this opening remark - what do you mean by "unknown", again?
Is it "isRequestFailed()" with unknown getIndex() code or what?
Mind you, roVideoPlayerEvent does not have "isUnknown()" predicate, so the choice of printing it to console as unknown was done by you/your developer. Mind troubleshooting that a little more? E.g. put a STOP in the catch-all branch that handles unrecognized events and when side-loaded and this happens, try investigating manually in console what that event is (getIndex(), getInfo(), getMessage()).

[RokuDale]

 7.7 build 4137 has an additional change to fix the streaming files for the continuous stream is larger than 4Gbytes.
Is anyone interested in trying this out? Just need a serial number of the roku to put it on our plex on 7.7 firmware test list.
thanks
rokudale