[velocity]

Porting is always several months getting depends sorted out  

Well, that definitely makes me feel better that this is taking so long. I've only had garbear's Steam Link port to compare it with (which isn't a fair comparison obviously).

Without a good depends, forget about tackling the main code base. Waste of time.

Hm, well yesterday I decided to only build the required dependencies for now. I should have probably done that from the start, because if I hit a major roadblock, then it'd all be for nothing.
So I only have 4 deps left, libdvd stuff and another one (required dependencies, minus the ones devkitPro already offers as portlibs, minus the ones that I've already gotten to build).

For a port to be successful, you need GL/GLES support and hardware video decode.

Switch has OpenGL support, the libnx guys wrote a mesa/nouveau driver a couple of months ago.

Without hardware video decode, any port is useless. That's what stalled the iOS/ATV2 porting for so long. There's no platform except desktop that has the ponies to software video decode the formats that are of interest.

We discussed this at DevCon, consensus was that Switch's Tegra X1 is quite capable and should be able to do 1080p H.264 decoding. Also ffmpeg is ported and also available as a portlib, so that's no issue.

About hardware support, I've asked around and people in the Switch scene are actively reversing its Nvidia NVDEC implementation (version differs from the disclosed public API) so actual hardware decoding should also be possible, eventually.

Good luck, I've been one of the main drivers in porting xbmc/kodi to iOS/tvOS and Android was well as a few other platforms that never saw the light of day. It's not trivial Be prepared for many rinse/repeat cycles and long, long nights of no forward progress.

Thanks for the advice!