> IPFS, or InterPlanetary File System, is a distributed file storage system that aims to replace HTTP. Both the Internet Archive and Neocities serve web content with IPFS.
Can someone explain to someone less smart (i.e. me) why IPFS and HTTP are mutually exclusive, as the above description seems to suggest?
No sure, I understand they are different protocols, but unless IPFS turns into a protocol capable of describing application semantics (which is what HTTP is) then I don't see how it can possibly replace HTTP – file transfer is not its main purpose after all. What I'm saying, I guess, is that I don't understand why HTTP and IPFS can't coexist quite peacefully? I.e. IPFS being a lower level transport protocol upon which HTTP provides application semantics. Or is that the aim? I don't fully understand.
They coexist already. IPFS just gives you a standardized way to store files in a global p2p content-addressable filesystem (or private content-addressable filesystems, if you don't connect your IPFS server to the global network). Imagine if you tried to visit news.ycombinator.com via HTTPS while visiting Mars; the timeouts would have to be quite large.
If instead the static assets, like html, javascript, comments, and stories are stored in IPFS, then you can load them from your local IPFS server node instead. You would probably even access that node via HTTP(S), and in principle the URL in your urlbar would be exactly the same as it is now. The rendering would all have to be done in JS though, which is a bit of a change. You would configure your local IPFS node to periodically update it's local cache so that the site is as up to date as you desire. (There's some handwaving there, but I guess you could just use a cronjob to request the content, forcing it to be cached.)
In practice, posting a story or a comment might have to look a rather different in that implementation. It would probably look like a throwback to UUCP or FidoNet, where distant nodes would contact each other regularly to upload messages. You would use PKI to discard messages from non-users, and linking the good ones into your branch of the filesystem so that viewers could see them.
Things like this that require logins are where most of the handwaving is with IPFS; the actual filesystem parts seem to work pretty well.
HTTP is a centralized protocol. IPFS is a decentralized protocol. It's considered a core requirement of the replacement protocol by these people. So, HTTP can't fit the bill.
Can someone explain to someone less smart (i.e. me) why IPFS and HTTP are mutually exclusive, as the above description seems to suggest?