Mastodon Feed: Post

Boosted by cstanhope@social.coop ("Your friendly 'net denizen"):
jonny@neuromatch.social ("jonny (good kind)") wrote:

oh-kay phew. some final cleanup, and then it's time to launch the alpha of this torrent tracker designed for rapid-response data rescue where a bunch of people each scrape chunks of a big thing and then recombine it later, where nobody can store the whole thing in one place because it's too big. then early next week we'll have the system for claiming and handling in-process scrapes. then we start fucking around with the fun stuff. has there ever been a torrent tracker that speaks activitypub to distribute uploads around between instances? because this one is gonna.

#HackThePlanet #MemoryHoleThis #YouCantKillAnIdea #ServersDieSwarmsSurvive #BittorrentIsStillSoGood #OverpoweredP2PProtocolEvenThoughIts20YearsOld

Attachments:

• video: [Screen recording of uploading a torrent that corresponds to a few parts of a larger dataset. The text throughout the example is unimportant, most of it is placeholder text that just has autogenerated corporate-sounding gobbeldygook in it, so i'll only describe the meaningful text] First create a single dataset part "A part" with a description "very special part" and a set of files that are contained within the part. Then create parts in bulk, specifying only a sluglike identifier, "second-part", "third-part" Then return to the list of parts that contains them, and a strip of checkboxes with 'upload' button above them. we check a few boxes and click upload, and a prompt to upload a torrent appears. It shows the parts that the upload will correspond to and a reminder to check the upload guide! We upload test.torrent, and the upload box expands, showing details about the torrent. the files are all from the Tentacoli (1977) OST. We give some more gibberish as the explanation of the scraping method and the upload, and hit upload. we then are shown a download page for what we just uploaded! all the information is there. We are able to click around through pages for the parts, the containing dataset, etc.