Just loaded my first #p2p site published to the new version of https://distributed.press with proper #ipfs #hypercore and #dnslink support!
The DNSLink stuff is a bit iffy for hypercore, but I'll be fixing that up tomorrow.
Now stuff can be published via Distributed Press, and viewed via @agregore
What's cool is that in addition to keeping your content online, you can easily configure it to set up DNS keys for you by delegating with an NS DNS entry for the `_dnslink` subdomain on your domain.
@bgtlover @agregore You can publish static assets only so if you want a DB you'll need to use sqlite with some JavaScript. We're working with sutty.ml for this grant which is a sort of Wordpress alternative which generates static websites and is going to have a toggle to enable publishing to Distributed Press directly. Docs are gonna come later this month. :)
@mauve @agregore how are you using ipfs without having to remember a hash of the folder you want to share? Ipns isn't much different, since you don't get a human friendly name to just type in a browser. Or, are you using those special fields in a dns txt record? I was planning to make a file storage+chatroom platform, like soulseek, but using activity pub for communication and ipfs for p2p storage. What deturred me from continuing is what I think is a weakness in ipfs, the fact that it became so decentralised, humans them selfs can't find information in there anymore. I mean, it could probably be attached to a blockchain where it's more common to have hashes around in blocks and such, so it won't be out of place there, but I can't imagine how it could be made more human-friendly without using centralisation, like a global nameserver for ipfs thing, which then defeats the whole point of using ipfs in the first place. So yeah, how did you work around this issue? or am I misunderstanding?
@bgtlover So, distributed.press embeds a go-ipfs node into itself and provides an HTTP API for uploading tarballs to publish.
For naming we use a combination of IPFS+IPNS+DNSLink. DP has a custom DNS server which you can point your domain at with `NS` records, then when a users tries to resolve your domain DP will serve DNSLink records which point to an IPNS pub key. That then resolves to an IPFS CID. With this updates happen at the IPNS key layer, but links stay the same.
@bgtlover This does mean relying on DNS infrastructure to do the name resolving. One thing I was thinking of is reaching out to Handshake to have an easy way to register domains on their chain and point them at a Distributed Press instance which can then resolve any DNS queries.
@mauve yep, that would have been my strategy for what I was trying to build, but at the time I considered it cheatting, since at that point I used more centralised technologies than decentralised ones, so what's the point? but yeah, this idea seemns more interesting indeed, even though probably not the way the ipfs creators imagined it being used. But yeah, who knows what the creators thought it being used for and what ways are there to do so, but it's good people are finding more interesting ways of using it.
@bgtlover To be honest, the DNS stuff is there to help pave a path from folks that are used to nice website URLs they can type in. For a more decentralized approach I think we should be using web of trust networks to discover people. Like, I send you a link or you scan my QR code and then you start seeing me and anyone I follow. Kinda like how the fediverse works now, but without needing a central instance. Check out https://cblgh.org/trustnet/ if you haven't yet.
@bgtlover For example you can check out `ipns://example2.mauve.moe` in Agregore and see it resolve to the raw file contents.