IPFS / HYPERCORE research

[despiegk]

• how easy to use with zerostor (new front end)
• how much work to be able to use 0-stor as backend for hypercore or ipfs

links

• https://github.com/hypercore-protocol/hypercore
• https://ipfs.io

[despiegk]

we should go non redundant mode and use our backend to make it redundant specs for the final server see https://github.com/threefoldtech/home/blob/master/specs/tfgrid_stor_server_p2p.md

[zaibon]

For IPFS, it seems the interface to implement is define there: https://github.com/ipfs/go-datastore#documentation

[zaibon]

I had a semi working POC of a 0-stor implementation of the datastore interface of IPFS: https://github.com/zaibon/go-ipfs/commit/3d8c4cd98893dfb463b512e581263494599e397c

[DylanVerstraete]

I started looking at both solutions. They look very alike, but are also very different.

IPFS pro's:

• has a great community
• the docs are amazing
• many implementations:
  • go-ipfs
  • js-ipfs
  • python ofcourse
• alot of client libraries..

The only downside I am reading of is that it misses a mutability feature, every time the content changes of some sort of file then the address changes as well.. It feels the same as changing the webservers IP each time it's content changes..

DAT / Hypercore pro's:

• mutability..
• easy to use

DAT / hypercore has, in comparaison, a tiny community, and AFAIK dev relies on very few core programmers. But DAT / Hypercore is Javascript only, it misses other implementations.. I don't think it will play nicely with zerostor.. Unless we create our own minimal implementation of it in golang.

It also has alot of nice projects being created on DAT, but they seem more like POC's than actual real stable projects..

• https://ara.one/
• https://www.wireline.io/
• https://beakerbrowser.com/ This browser is really cool! Everyone should take a look at this!

[zaibon]

will be continued at https://github.com/threefoldtech/home/issues/842