• 691 Posts
Joined vor 7 Monaten
Cake day: Apr. 18, 2022


Sorry, I mean, given a text get the hashtags appropriate to use with the text.

Describing my perfect p2p file-sharing program
cross-posted from: https://lemmy.ml/post/622699 > The program would have monetization because most programs without it die sooner or later. People would be able to download a few files per day, if they want to download more they need a subscription or they need to keep a share ratio of 1 or higher. > > The program would be connected to a server through onion routing. The server would have a collaborative database of file names, ratings, etc. It could be used to show the agreed upon name for each file instead of the name each user has given it. It would also provide content suggestion based on the user ratings. It would also have collections. So that people would be able to wishlist a collection instead of individual files and the collection would be downloaded automatically, as the files become available, based on the user preferences, for resolution, bitrate or whatever. People could also wishlist individual files. > > Then it would have the typical search where instead of searching the server you are searching for currently available files and folders. Fuzzy search with advanced options like: > > - exact match: `"Keyword1 keyword2"` > - exclude: `-keyword` > - or: `"|"` in between keywords. > - `*` parameter to match anything. Like some web series `s1ep*` > > Greater than, less than, or a combination of both for: > - size > - date (age). > - number of files (in folders) > - number of seeders > > You add files and they are scanned without hashing so people can download directly from you. After they are scanned they are hashed so that the network has swarming. The more people with a file the faster it downloads. I think this should be optional and activated by default, because hashing too many files takes too long. > > It would allow searching for files and directories. It would allow browsing users shared directories for content discovery. > > I want to have a clearer idea of what features this program could have. Contribute your own ideas.

The maintainer is shutting the server down. He was looking for someone else to maintain the project and I don’t think he found anybody. When I used it, it didn’t connect because the bootstrap server was down and he had to turn it on. So going forward, and without no-one doing that, there’s probably no way to connect to the network. Also I didn’t found anything when I used it. It’s I2P over gnutella network and has swarming, but I guess it’s slow because of the onion routing and it didn’t gain traction.

3 stars. It leaks memory. If you queue more than 100 or so files for download it closes randomly, even with 16GB ram. Download is very slow. You can only search for single files. Can’t browse most user shares either. Good for finding old/rare media.

1 star. You can only search for single files. Can’t browse users folders either. I’ve tried to add a folder but it didn’t find any files. I haven’t used it after that. Has swarming? I don’t know

0 stars. It’s supposed to work on wine, I could run it but couldn’t share anything. I think this would be my favorite if it was native and worked.

3 stars. Only for hit & run. Seeding there promotes it’s usage which is something I don’t want to do. It fragments the network by not sharing peers across torrents with the same files. Best to find files less than a month old, otherwise it won’t be seeded unless it’s popular.

3 stars. At first I shared all 70TB and connected to all hubs but then the program didn’t open. After deleting the configuration I added only a few files and connected to 10 hubs and it works fine. Most of the people there speak russian and browsing their folders is impossible without understanding it. But there is a search for directories filter so you can always find the right folder in english that way. What I don’t like is that there are around 40k users but they are divided in around 300 hubs. To make sure no-one has something you are looking for you would have to connect and disconnect from a few hubs at a time while searching the same thing over and over, which isn’t practical. So file discovery is good but it could be better if you weren’t limited by the number of people you can connect to. Has swarming? I don’t know.

4 stars. Only allows searching for single files, but you can browse the users folder so it has good content discovery. It’s the only place where I can share all files without problem. No swarming. There’s mostly music.

5 stars. It’s closed source and there aren’t many people. It took too long to open when I shared all 2M files but after removing some it works fine with 500k files. You can browse users folders. Has swarming.

Is using a USB drive as swap partition a good idea?
To avoid writing so much on my SSD I've moved the swap partition to a 32 GB drive which goes for $7 nowadays and if it dies it won't hurt unlike if the SSD dies. I've also lowered the swappiness from 60 to 10. I assumed it would be faster on a USB than on the hard drive. Was this a good idea?

Orange Pi 5: 8-core CPU 2.4GHz, up to 32GB DDR4
I've looked at the 4, 8 and 16 GB and they are at €72, 90, 126 with preorder during a few days.

Yes but with more than one hashtag and with compatibility percentages.

I think there should be many algorithms to choose from or some way of configuring the algorithm on a per user basis so that you don’t have to make your own instance if you want to change what the algorithm shows.

I saw yesterday someone who had to select the pictures in which he appeared from a captcha. One of the pictures seemed to be from another person where he appeared looking outside his house's window. But now I don't know where I saw it.