feat(hls): add keySystemsMapping to hls config

[avelad]

Add new config manifest.hls.keySystemsMapping to allow use another keySystems. Por example: com.microsoft.playready.recommendation

Closes #4243

[joeyparrish]

I am not convinced that this is the best solution for com.microsoft.playready.recommendation. I believe that for some devices, PlayReady must use the older key system ID. In MS Edge, you can use either, but should use the new one.

I think we might need some kind of built-in fallback in DrmEngine to automatically try and prefer .recommendation when possible.

[avelad]

We have the same solution for DASH

[joeyparrish]

We have the same solution for DASH

I had no idea! I did some digging and found PR #3276, which I even commented on in an approving manner before it was ultimately merged by someone else. No memory of this! 2021 was a rough year.

I would prefer that you move this field to the manifest level and combine with the DASH version. You can see an example of a backward compatibility shim here:

ac5acc8#diff-7abf4aa2639707ce10a19703663841fae6c5638809faf4dfb2f1894791f154ddL2948

If you do something like that, we can move the field now, but don't have to break compatibility until we're ready for v5.

[avelad]

Ok, I’ll do it tomorrow. One question, are you agree with my config? Or do you prefer another type of mapping?

[wjywbs]

I think we might need some kind of built-in fallback in DrmEngine to automatically try and prefer .recommendation when possible.

I do see the value of a default mapping and automatic probing equivalent key system names in an order. By the way, I started a discussion in Chromium for supporting PlayReady key system in Windows:

https://bugs.chromium.org/p/chromium/issues/detail?id=1328921

[joeyparrish]

@alvaro, yes, I think this is fine. I would have preferred it if we didn't require a configuration to use the best key system ID for PlayReady, because I think many people will overlook this. But it's also simple, and we're already doing it.

Although @valotvince said before that his attempts to do an automatic failover were turning out to be too complex, I would still be interested in something like that in the future. It wouldn't have to be in conflict with the mapping necessarily, and perhaps both could coexist.

[avelad]

Replaced with #4254