I’ve recently run into being fed up with the technology that is domains. I’m looking into sharing my content as magnet or ipfs links, but I see they aren’t supported here. Why? http/s clients are needed (and only the most simplest ones) to access lobste.rs but simply because lobste.rs is a centralized aggregation service which is free to pick the protocol it wants to use to sync readers. It takes the place of an RSS feed or similar, but the RSS feed can at least share any type of links while lobste.rs can’t.
I just don’t think we should expect a HTTP(S) link aggregator/discussion forum to link to anything that requires third-party software to run. We don’t do .onion links here either, and I don’t even think sites with opennic TLDs, for example, would be appropriate.
While I can kind of agree with your sentiment, this is just not the forum for this kind of thing, in my view.
Maybe I don’t understand this enough and could use some education, but don’t magnet and IPFS links need custom app support? I primarily browse on my phone and anything that needs more than my local apps would be a huge deterrence.
My only experience with IPFS has been to get libgen.rs to work on my phone but I could not figure that out at all.
For what it’s worth, Gopher and Gemini links are already supported.
Only in the homepage field on user profiles, not as story links.
Oh, right!
Some browsers include IPFS support, most notably Opera. And in Firefox is easy to add support with an extension. Probably in other browsers too, because an easy implementation can be made leveraging the IPFS work to a gateway.
In the end, mainstream browsers do not include IPFS support because there is a very small number of users, but it’s a vicious circle, because if browsers don’t include support for it, many devs do not consider to use it (like here in Lobsters)
Probably cause people can’t see content distributed via IPFS or bittorrent without special clients. The argument that HTTP/s content also needs a client is moot cause to access Lobsters itself you already have a https client anyway and also most devices these days include a web browser.
This is why. When CanIUse shows that popular browsers mostly support these links, Lobsters will permit them.
Discussions where we’re just riffing on the headlines tend to be pretty bad. Either we have no comments or we have people rehashing whatever favorite argument shares a few keywords with the title. This is why we find canonical links; swap in archived texts when sites are down temporarily; have dozens of lines of code special-casing resubmitted links; put our limited volunteer time to fixing broken links; and remove links that are paywalled, lead magnets, or sales pages for otherwise topical books/courses/etc. If the link doesn’t work, the conversation doesn’t work.
I’m sympathetic to the chicken and egg problem here, that in a multiparty ecosystem of browsers, web hosts, network operators, and sites like ours, no party wants to spend their limited resources supporting something that won’t work until some critical mass of the other parties also do, with the result that nobody supports it and a good change is lost. But we’re the least able to influence the others or risk a change like this. For the other parties, the cost of supporting a protocol that goes unused is a rounding error on an engineering budget. For Lobsters it’s a small code change, yes, but we’d be shooting holes in our community with broken links and bad conversations.
Lobsters is a community of people, not “a centralized aggregation service”. Lobsters is far more than a couple thousand lines of code and a database, it’s the gut feeling that it’s worth your time to participate because other interesting people also feel it’s worth their time to participate. Communities are enormously sensitive to feedback loops, and this proposal to allow links that don’t work threatens the primary one that brings people back to this one.
Thank you very much for the articulated response and reasoning! :) I guess this segways into building my own aggregator ;) You’re 100% right that lobste.rs is both its community and the service itself! (And thanks as always for maintaining it!!)
And for those wondering, magnet link scheme support is only at 30%Nevermind, this is just to register a handler…: https://caniuse.com/?search=scheme%20magnetI don’t think the ability to register a protocol handler for
magnet:is fair to count as “support” for this purpose.scheme:ipfs is similar and no, I wouldn’t consider either a working link for our purposes.
Right right, they really need a “native support” thing on CanIUse for magnet…
curiously, I whitelisted ipfs as a potential valid protocol schema in Firefox seven years ago. At that time, I thought that would help us move all those protocols into the browser. It didn’t happened.
As a reader of lobste.rs browsing /recent/, whenever I see a submitted link that is difficult to open, I just skip to the next submission. Slow webpages, obnoxious banners, prompts to “view in the app” or, in this case, a fundamental requirement to both use an external app and leak my IP to an unknown amount of people before I even get a chance to understand what the submission is about. It’s all the same to me. If there are that obvious hurdles to reading something, I just assume it wasn’t that important in the first place.
Sometimes a link gets upvoted far enough to make me care about the content regardless of its presentation. But I’m still not sure we should allow magnet-links just on the off-chance that might happen.
When I post Gemini content here, I do so through a Web proxy link.
Everyone has a browser installed that can work with the internet as we know it, via https. Not everyone has whatever you need to read a magnet link.
As a general principle, making content accessible to the most amount of people is a good idea. As a principle for a site dedicated to in-depth discussion and knowledge sharing, accessible content is even more important. Otherwise, a section of the audience is excluded.
The same discussion comes up a lot in multi-lingual contexts. If you have 3 Norwegian speakers in a group of 5 people, and those 3 start talking in Norwegian to each other, then the other 2 are left out, harming the overall group collective.
Even worse if the Norwegians are split between nynorsk and boknorsk!
this is actually a fun idea. If no one else has done it yet, I will. gimme a month or so to slide it into the project queue.
Seems like it might be a better use of the time to implement them on browsers, so we can all use them
You could use just the ip if the problem is the domains :)
Magnet and IPFS links are not about the domains, it’s that the file content is distributed in multiple computers, so you could download each chunk from a different server. And the original server doesn’t need to exist anymore if there’s enough replication in the network.
OP was complaining about the technology in domains. Without more context, I’d assume the problems of having to buy one (with ransomy prices for some), DNS (with impersonation and such) or something like that. There was no mention of distribution, decentralisation or serverless. You should be able to download from multiple servers with http domains, using multiple A records, no dns caching and range. Replication is normally done with anycast on the IP level, or archives. I guess this is both serious and not-serious answer at the same time. And again, we don’t know what was fedding up the poster
IP ownership is even worse than domain
Maybe, I’m not too much into that part. Do magnet or ipfs solve issues with ownership?
Content addressing is inherently ownerless, but with IPNS the locator is a public key pointing to a specific content hash, and whoever has the private key can publish an update. I don’t know how well this ended up handling propagation and staleness in practice, since I haven’t been up on IPFS for years and I couldn’t get it working well in my tests.
magnet torrent links could do something similar if BEP-46 mutable torrents is usable, but that looks like it never got the DHT support needed to implement, dunno why. Maybe BTv2 has an answer for this?
No, it works well, and I’ve used it in production systems.
BT V2 makes it a bit nicer but isn’t a requirement.
See https://github.com/anacrolix/btlink too.
[Comment removed by author]