What do you guys think about IPFS? Will it ever go anywhere? Is it junk?

I think it's also important to be able to specify fault-tolerance characteristics. For example, a file might be specified to have at least three accessible copies, each in geographically distinct areas of the world. An economic system is implied in all of this, and that, of course, needs some deep thought. Information economics is inherently different from the other economic systems - industrial, kingdom/mafia/gang, agrarian, hunter/gatherer. Of course, the banking/mafioso types will try to weasel their scams into this new arena. Can you think of a few distributed file system examples where this is the case?

How about chain of custody on a file? A high integrity distributed file system would need to be able to provide reasonable assurance that a file has not been spoofed or modified. Any derivatives of the file should also be auditable.

>what is stored,
You control what your node stores.
>what can be accessed
You decide what to put up
>who can access
Storing private content is not the use case of IPFS but even for that it's no harder than anything else, just put it up in an encrypted archive and send the password to the people you want to retrieve it.
>network data rates per share per person
The node doesn't currently have integrated rate control and it's very annoying.

It sounds like you are describing Information Assurance (IA). en.wikipedia.org/wiki/Information_assurance

>A high integrity distributed file system would need to be able to provide reasonable assurance that a file has not been spoofed or modified.
That is like the one thing that IPFS actually does, files are identified by their hashes so the file cannot be modified more or less by definition.

The last time I tinkered with it, FUSE was so slow and unreliable that it was basically unusable; the system load due to IPFS was enormous; it seemed like it was necessary to keep two copies of everything but I don't remember why, maybe the FUSE problem; network capacity is unlimited and free - at the time I evaluated IPFS, I had a monthly cap.

A complete Information Assurance system would need to be quite a bit more elaborate than just identifying a file by its hash. For example, the IA system might need to provide chain-of-custody guarantees on the file hashes...

>capacity is unlimited
...capacity *isn't* unlimited...

What exactly are you trying to assure? If you're trying to assure the file you're viewing is the file you were linked to the hashes do exactly that. If you want to assure the file you're viewing was created by a particular individual that's what everyday PGP cryptographic signatures are for.

It needs the ability to let other people added content in a limited fashion to your "site" so that content discovery is possible without relying on centralized side channels.
It needs IPNS to not be the slowest thing in the universe.
It needs a human readable decentralized naming system that isn't piggybacking off regular DNS.
I don't know if it will take off after those things are fixed, but it certainly won't take off before then.

add content*

>literally just rebranded torrents

It annoys me that Dat is a more interesting protocol, but their primary implementation is in fucking JavaScript.

I think a distributed file system needs capabilities that are not present in IPFS, and adding these capabilities on top of IPFS would be incongruous with its fundamental architecture.

What capabilities are you picturing exactly?

If a high integrity information system is to be based on a distributed file system then that DFS needs many deep assurances.

>Can you think of a few distributed file system examples where this is the case?
LOL
techcrunch.com/2017/08/10/filecoins-ico-opens-today-for-accredited-investors-after-raising-52m-from-advisers/

You don't actually know, do you? We're two for two on me asking "what features do you want?" and you replying with vague allusions.

>Generic comment about how finished decentralized networks are if people just post cp to it
>Well reasoned comment about the functionality and purpose of IPFS itself
>Ad hom reply

Every. Single. Thread.

On the flip side...
Any news on what happened to the C++ interface? It just stopped being developed.

Do you really expect to develop the requirements and specifications of a distributed file system in a Jow Forums thread on Jow Forums? It is amazing this conversation, as it is, took place.

Not even him but yes.

Use your text and convey your thoughts to my brain.

Attached: 1539153400752.gif (360x238, 1.43M)

dis be a ipfs thread, yo! talk bout c++ and gentooo or gtfo!

based

I expect him to clarify what he means by distributed file system, IPFS already claims to be one but clearly that's not what he means.
Is he picturing IPFS but with signatures and private file encryption built straight in and it somehow magically records who has and hasn't accessed a file?
Is he picturing something federated with file servers connected into a bigger network and a more traditional architecture otherwise?
Is he picturing something blockchain-esque where the right to modify the authoritative version of files has to be transferred between users represented by public keys and changes to that version are tracked as items added to the distributed ledger?
I think it's probably the last one but I'm really not sure.

Firstly, I don't think a distributed file system necessarily needs to be completely decentralized. A centrally organized, distributed architecture, while less idealistic, might be more manageable and practical. Also, I don't think it necessarily needs to be universal or global, federations might make more sense.

It's blatantly superior to http.
The feds go after whoever is seeding it.

Federations are a very solid middle ground so I'd be interested in anything that used them.

time for a shower.

Attached: shower.jpg (210x240, 13K)

>I think it's probably the last one but I'm really not sure.
A blockchain isn't necessarily the best way to implement a trusted database, IMO.

I'm a little dubious of that as there is a billion dollars to be made for the first person to come up with a better model for a distributed trustable database that's better than blockchain. (Unless you're referring to HashGraph or one of the other already deployed technologies in which case fair enough).

>currently two "let's install gentoo" threads
Jow Forumsanitor's Jow Forumsang-tards are hard at work...

it will go to mars and we will leave behind all niggers and kikes

If it runs on top of the internet how is it a competing service?

>distributed trustable database
I suspect most of the effort is directed towards decentralized systems that run on the current infrastructure. If it is a {"there is a billion dollars to be made"} type of problem (I think it is) then maybe it is worthwhile to revisit core Information System technology and consider the design of a foundation more suitable to this particular problem.

Wow, this thread hasn't been deleted yet? It is *so* not status quo. The Jow Forumsanitard must be AFK.

>not status quo -> delete
that's a common british style of tardiness, isn't it? hmm...

>It needs IPNS to not be the slowest thing in the universe.
How in the world is a hash lookup of a single public key string so time consuming
AHHHHHHHHHHHHHHHHH

Why is IPNS so damn slow?

Attached: aaaaaaaaaaaaaaaaaaaaaaaaaa.jpg (457x309, 97K)

it competes with other services like torrents, tor, freenet and god knows what else. ipfs is not a competition to internet lmao

freenet, but it doesnt guarantee the file will be accessible if no one uses it for a long time