Not regularly, but if you're looking for a super obscure documentary that is not on youtube and you cannot buy anywhere - emule/amule may be the place to find it if you're willing to wait a few days or even weeks. Did that a few times.
DC++ used to have the most obscure stuff. I remember finding one site(?server?hub?) that had every single F1 race broadcast (practice, qualifying, and race) for many years. Basically someone had been recording every race on VHS tape and eventually digitized them all and uploaded them. This is just one example of the kind of specializing you could find. I wonder how much of it is still there.
One of the reasons why I like IPFS is that it gives you an immutable links to files like these. If you find the file once, you can get the link and access it again later (assuming people are still seeding) and be sure you're accessing the exact same file you think you are.
I mean, magnet links are the same thing—with the exception that they aren't widely used between users, to my knowledge, i.e. people aren't too familiar with them and go to search sites instead. (Not to say that the same won't happen in IPFS if it ever gets popular.)
They're not the same thing which is why they're not as useful.
Magnet links apply to an entire collection of files, as snapshotted by the creator. There is no way to pick a single file and refer to it, immutably. It's all or nothing. Additionally, metadata besides the file contents will change the hash. This includes filenames, directory structure and piece length (chosen at snapshot time, by the creator).
All of these limitations are enough in practice to make magnets useless for widespread per-file distribution.
Ah, if it's the entire dictionary and not filtered by the clients to just those fields, then presumably additional data may be shoved in—specifically the trackers to contact.
No, look at the content of the dictionary: trackers are not part of the "info" dict, so you can add all the trackers you want, it won't change the identity of the torrent. Identity is defined by the infohash.
Not sure if any metadata other than the file name affects the hash—e.g. modification times. Chunk sizes likely do. Otherwise, I'd expect the hash to address actual content, though frankly I haven't checked.
Edit: in fact, my hasty, belated and superficial skimming suggests that IPFS is an implementation of just a DHT (presumably with some metadata and chunking built in)—i.e. it keeps data itself in about the same way that Bittorrent stores torrent descriptions in the Mainline DHT.
So I tried a few years ago. Was looking for an old obscure TV show and couldn't find that anywhere. Installed aMule on a whim, got it connected and used the search function. Actually found the show, started the download. A minute in it found a couple peers and started downloading, very slowly. Then I remembered that you can see the file name that all the other peers stored the file as locally. It was handy back in the day (around 2001) to spot fakes. For some reason, people back then liked to rename random movies or porn to current blockbusters, so when you look for a new movie you get the wrong thing. That's why eventually websites that manually indexed links took over.
So I opened up the window that shows all the file names of the tv show I just added to the download queue, and most of them very explicitly hinted at child porn. I freaked the hell out and nuked the whole installation, deleted everything, overwrote empty parts of the disk with zeros. It just downloaded a few kilobytes of a couple hundred MB, but still... I just thought what would have happened if I hadn't remembered that file name thing and would have found out when trying to play the damn file after downloading it. So probably not gonna use it again anytime soon.
> I freaked the hell out and nuked the whole installation
when something so bad and rare is used as scapegoat for so many unrelated things as part of a fascist power grab, that everyone rather run away than to report when the actual crime happens to the authorities?
makes you wonder how much all the 'think of the children' knee jerk legislation actually helps anyone, besides the politicians
I appreciate your view on the matter. In this case, however, as disgusting as it sounds, it's the "authorities" themselves serving that kind of content. Actual honeypots. No point in reporting, they perfectly know. Better to just run away and pretend nothing was seen.
In fact, it is arguably _more_ useful than BitTorrent on the majority of files people might share. You see, with BitTorrent, you typically share few files, and mostly the ones you just downloaded. There is little incentive to actively seed thousands or tens of thousands of files you don't even know are interesting to people.
With the ED2K protocol, the premise is that you share a lot: complete directory structures, and you search your many hashes when others make a request. That is more or less _the_ way to share collections of items/files/media/etc (other than, perhaps, running your own tracker which shares every file in your sharing directories separately).
I occasionally use aMule on ubuntu (sudo apt-get install amule). Mainly to download classic Italian movies. There are a lot of titles to choose from Commedia all'italiana and neorrealismo, it's fast and I don't have to visit any shady webpage to get them, I just use aMule's search bar.
I haven't used EMule in a long time either, but I still use Soulseek occasionally to find stuff that just isn't available anywhere else (legal or otherwise). I have had wishlist searches in Soulseek that were active for months before someone logged in who had that particular thing.
