Support only built-in OS decoders/encoders. Remove any GPL code.

[EvanRuiz]

Thanks for putting together this great project!

I've been evaluating using PopH264 and am investigating the legal/license aspects. It appears that as long as I don't distribute any code that does the actual H264 decoding/encoding, I would not be liable for patent licenses to MPEG-LA in any scenario.

That implies if there is a hardware-accelerated only PopH264, where only built-in decoders/encoders are used, then that would be in the all-clear.

Do you have any interest in officially separating the project out clearly by license and patent-free aspects? What do you think of this?

I also noticed the software fallbacks of Broadway and X264 have their own licenses, and I don't think those can convert to just PopH264's Mozilla Public License Version 2.0 license without at minimum an attribution by anyone using PopH264. People may be missing this attribution requirement if it isn't clear in PopH264's license and/or readme.

X264 is GPL which I think is particularly problematic, and its inclusion possibly should make PopH264's current project license GPL and any projects distributing/linking X264 turned to GPL too unfortunately.

In summary perhaps there could be: (names a placeholder, and defer to your preference)

• PopH264.HardwareOnly (Or just PopH264)
  • Patent-Free
  • No X264, No Broadway
  • Licensed w/ your current choice of Mozilla Public License Version 2.0
  • SoyLib MIT license looks like it needs attribution in main project readme or license file
• PopH264.SoftwareFallBacks
  • Not patent-free
  • Fees may be due to MPEG-LA in some scenarios
  • Licensed GPL as long as X264 is included
  • Each include project such as Broadway needs attribution credit in main project readme or license file

I may start working on the 'PopH264.HardwareOnly' part as a PopH264 fork for my own purposes - and happy to contribute it back officially. This issue is intended to start the discussion on what the official project plan should be.

[SoylentGraham]

At this point, I would probably just strip out X264 & broadway entirely. Now we have WebCodec supported in a lot of places, there's little need for them in emscripten, and broadway has such a low profile support (baseline3?) it's little use.

They can be stripped essentially with the ENABLE_XYZ defines, but should be little effort to cut from the repos.
Submit as a PR!

Soylib is also being stripped out, as c++17/20 started replacing a lot of my ~2010 classes :)
being MIT, i dont think it needs attribution? (Im not always great with the legal stuff - but its my lib anyway :)

FYI built in OS decoders doesn't mean hardware, so I certainly wouldn't name anything thus.

I don't think there's value in forking the project, its lasted this long by being simple, and (if I had a need) i'd be simplifying it further, I have a few libs which have slightly improved APIs etc. (poph264 is still in active use, just hasn't needed any updates :)

in summary, feel free to submit a PR stripping out GPL libs!
Better than forking the project into two things.
If anyone ever wants those two components back, they'd be easy to re-add, or make another lib with essentially an identical API

[EvanRuiz]

Thanks for the guidance @SoylentGraham !

I will work on the PR and submit for your feedback.

[EvanRuiz]

I've updated the title @SoylentGraham to now describe the intent of the PR more accurately, please re-edit as you see fit.