https://github.com/AquiGorka/test-ipfs-file-sharing
I understand there are a couple of issues around that might relate to this, in that case this is just double checking you are aware of this behaviour and asking for an ETA鈥揑 did try to hack around to make this work modifying the MAX_MESSAGE_SIZE as discussed here but that did not work.
By "big" I mean: not able to share files bigger than 500Kb (actual limit might be different).
I've tested this with https://github.com/ipfs/js-ipfs/tree/master/examples/exchange-files-in-browser multiple times and I've been always successful. @AquiGorka can you check again if this is still a issue was it ever a issue?
@diasdavid thanks for the reply, it is an issue.
I wrote a more detailed explanation here: https://github.com/AquiGorka/test-ipfs-file-sharing
There are two configurations for the signaling server:
webocketwebrtcThe example you mention uses websocket:
config: {
Addresses: {
Swarm: ['/dns4/ws-star.discovery.libp2p.io/tcp/443/wss/p2p-websocket-star']
}
}
which will always work well. In my test scenario if the signaling server is set to use webrtc the problems with "big files" appear.
My test uses a "private" IPFS network where the only nodes connecting are the browser nodes, I am aware there are some implementations off cli/daemon nodes to help transfer data.
I developed a GIF sharing app that uses IPFS and implemented cross-browser e2e tests, if the settings use webrtc (either local or IPFS) the tests fail, when using local websocket the tests always pass and when using the IPFS websocket signaling server, the tests sometimes pass and other times they don't (could never tell why, maybe a traffic throttle for connections on the signaling server).
You can check out the app & tests here:
@jacobheun this is the issue I was telling you about after your talk at IPFS camp.
@AquiGorka thank for pinging me here. I'll be looking into this as a part of https://github.com/ipfs/js-ipfs/issues/2161#issuecomment-502203711, as I believe the underlying issue has to do with a bug in Chrome and the max 64k webrtc message/chunk size.
Most helpful comment
@AquiGorka thank for pinging me here. I'll be looking into this as a part of https://github.com/ipfs/js-ipfs/issues/2161#issuecomment-502203711, as I believe the underlying issue has to do with a bug in Chrome and the max 64k webrtc message/chunk size.