Libp2p mobile client

[farhoud]

Our mobile client rn-fula it not ideal for production it has some deficiencies:

• It's run in headless browser which is another point of failure.
• It is hard to develop and debug! (passing error and log between react-native <---> webview <---> libp2p is possible only by sending message's)
• DEX is awful.
• WebView does not have access to disk so we have to read the file and pass it as stream of base64 to webview! (expo-media-lib will load/save file from/to disk <---> fula-client encode result to base64/binary <---> fula-bridge encode result to binary/base64 <---> fula-client stream data )
• Background job is now impossible because of the webview provider --> need to write our own webview for ios and android.
• graph subscription make it more complicated than already is.

I'm looking into alternative:

• run js-libp2p in react
  • failed, and if it work still to hacky for production.
• bind go-libp2p with gomobile in react natite

[farhoud]

I will try to bind https://github.com/libp2p/go-libp2p/tree/master/examples/echo in react-native sample and report my result

[farhoud]

@gitaaron @keyvan-m-sadeghi @ehsan6sha @masih These are my facts for migrating to go-libp2p I know it add complexity of writing go protocols for client, but i think that is better than develop and maintain:

• headless WebView without provider for background service (android and ios)
• write and maintain a protocol for bridge between webview and fula-client with all problem i mention in this

[farhoud]

I don't add any think around the performance because i don't have data. but a native code most have better performance than running a browser and then run js-libp2p on it.

[gitaaron]

Nice writeup @farhoud - as you mentioned - my main concern is having to marshal the data to a string to pass it back and forth between the webview <-> native.

[gitaaron]

As discussed in yesterday's meeting here are a list of pros/cons for going with webview / webrtc :

• pros - not have to maintain two versions of client protocol
• cons
  • inefficient (both webrtc + webview)
  • complexity of webview setup
    • webrtc + crypto deps
    • still needs interface to react native

[ehsan6sha]

I think here efficiency is priority 1