Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
PeerTube 4 RC (framablog.org)
272 points by booteille on Nov 30, 2021 | hide | past | favorite | 81 comments


The PeerTube developer community is super friendly and helpful to new developers. I contributed some minor improvements around a year ago, minimally touching several sides of the projects:

- documentation for the SAML plugin

- translation to my native language

- some improvements to enable better encoding performance through hardware encoding

My experience with the maintainers was nothing but positive, in all three areas. Such a warm welcome is rare in development land, so I commend the PeerTube maintainers not only for their excellent software, but also for their welcoming attitude.


That's nice to hear, sometimes my bugreports to open source projects was threatened like enemy attacks.


PSA:

If your are de-googled and use NewPipe to watch YouTube, it already supports PeerTube (and more).

https://f-droid.org/en/packages/org.schabi.newpipe/


NewPipe is such a nice piece of FOSS. Too bad that they decided against integrating SponsorBlock.


This fork integrates sponsorblock, been using it for a while and even though the maintainer calls the integration "rather basic" it works quite well. https://github.com/polymorphicshade/NewPipe


I'm really glad YT Vanced did integrate it.


It would be weird if it didn’t considering its name is “Youtube Advanced without the ‘Ad’” :)


What's the difference between Vanced and NewPipe?


Vanced is a mod of the stock YT app, NewPipe is a completely new codebase that doesn't use YT's API.


How does it get the YT data if it doesn't use its API?


The way your browser can watch YT.


Newpipe is Libre Software, Vanced is not


Yeah, just read through the GitHub issues thread and it's rather flimsy justification for not implementing it that seems to boil down to "because I don't want to". A real shame, since I'd always admired the project.


"Because I don't want to" is a perfectly valid reason to not implement something in free software.


I made an android client specifically for peertube also open source: https://github.com/sschueller/peertube-android

Any help with development is very welcome as I have little time to invest at the moment.


This project needs marketing people on the team.

I’m interested in peertube it’s an exciting concept. When I when to that page I expected a simple, clean design inviting me to download it or use it or whatever. Instead there’s a rambling unfocused page without a call to action.

YouTube is the Mike Tyson of video…. you’ve gotta be much better than this to get in the ring with YouTube.


The people behind PeerTube are the non-for-profit Framasoft.

Actually, there is only one developer, not even working full time, on the project, and some lovely contributors.

So Framasoft is actually not playing with same means than big players like YouTube or so. That's why they are not handling the problem the same way.

The goal here is mainly to build a tool for people and small structures. There is no intent to be the new YouTube.

The message is more humble: anyone should be able to host and share at low cost their own videos. That's the issue that PeerTube solves right now.

If you think all of that sounds good, think about donating to Framasoft. They are living mostly from donations and that's what made PeerTube possible.

https://framasoft.org/en/#support


The url points to a blogpost at Framasoft with information on the release canidate. If you visit the homepage of Peertube (https://joinpeertube.org) you'll likely find the information you are looking for.


I think this is a good example of what separates HN users from normal users. The fact that the blog is on a completely different domain and using a different name is confusing. What's more, when you visit the Peertube domain it's not clear what you should do. Do I need to download something? What's an instance? How come the top video on the page brings me to a separate page?

I know these are things that most HN users are happy to dig into and figure out, but currently there's no way Peertube is "an alternative to video platforms" for even saavy web users, let alone regular users. Maybe that's OK though?


not really, Framasoft is a French company that develops a collection of open source alternatives to popular software.

It's the same as if Microsoft was posting a blog on its own msdn.com domain for something related to visual studio or SharePoint


Small correction, it's a non-profit association that can be placed in the same cluster as Mozilla, the EFF, the CCC or the Linux Foundation (i.e., the good guys if you care about FOSS and an open software and internet world)


> It's the same as if Microsoft was posting a blog on its own msdn.com domain for something related to visual studio or SharePoint

Which underlines the point that the parent is making. Normal users are not the ones reading things related to SharePoint. Apple does their announcements on the apple.com domain and not on applnews.com.

> I think this is a good example of what separates HN users from normal users.


Normal users are also not the ones primarily expected to read release notes for release candidates for the Peertube server software. Sharepoint is arguably a good comparison: run by administrators for users - it's just that peertube is more interesting to nerds that are both in one (although I'm sure there's some folks with sharepoint at home...).


> Normal users are not the ones reading things related to SharePoint.

Normal users would never ever find this blog post in the first place. You are trying to fix a problem that does not exist.


The "Read the blog post" link is placed in the first highlighted box of the project's homepage :)


> msdn.com

Doesn't exist any more. It's now all under microsoft.com (e.g. devblogs.microsoft.com).


The way peertube is an alternative to youtube is not by end users understanding what it is, but by technical users hosting themselves the videos. For example, OCaml has a peertube instance to watch OCaml related content: https://watch.ocaml.org/. This is an alternative to watching the content on youtube.


I think this is the challenge with all decentralised systems, its not as easy as you click some buttons and you get served like in web 2. Hope more focus will come on experience part of this, for such approaches to become popular


If you go to that site you don't get tons of videos, you get a technical explanation of what it is and a link to a page with 10 channels. Unless you're the kind of person who is on HN, you're just gonna go to youtube instead.


Agree, the way it’s presented will pretty much guarantee nobody but hardcore nerds will ever use it, it’s sad that noble efforts hobble themselves in this way.

I mean isn’t it ironic that PeerTube doesn’t use a PeerTube video to introduce its new features? Guess their team thinks their own product isn’t a good way of consuming interesting information…


It'd certainly be nice (although an oxymoron) if Framasoft could singlehandedly create a deep decentralised network of video content creators to challenge YouTube. But that isn't really necessary - if all they do is bring down the cost of someone else to compete then that is more than enough to be helpful.

And while I do agree that a savvy marketer would have had a video in the article ... a video is actually a bad medium for communicating a new release. Videos are better for stuff with a bit of visual spectacle.


Does the license allow commercial use without releasing any modifications? Else it doesn't make sense for a potential competitor to provide improvements to create additional competitors for themselves.


I think you are still under the impression that success is still based on codebase and you just need the perfect source code to build the most successful service. That's wrong. Success is built on relentless labor, continuous marketing and luck. That's why Gitlab can build a successful company while dailymotion can fail to build what is functionally not new.


I think you're lacking enough data points to determine what impression I'm under, making assumptions as to my perspective, where my question is coming from.

There is a competitive advantage to not giving your future competition a heads up by providing them your keys to the foundation of your Kingdom. Sure, they'll copy what you do eventually if what you're doing is working, but why make that easier for them?

If someone is seriously going to take on the current systems and sees value in part or all of PeerTube then it's not a substantial part of the cost, because as you say, it's not the codebase that's important - though it becomes a conversation piece if it means any additions to the codebase reduces an even temporary competitive advantage from potential future competitors.


PeerTube itself is licenced under the AGPL.

But that wouldn't stop a non-AGPL licenced solution from interoperating as the network is based on an open standard (ActivityPub).


So the site is about a server you can run that will be your very own tube site. Who other than nerds is going to install that? It requires administration, you're going to be very interested in running an instance to the point of commitment before you go there.


Great idea and totally agree. Even someone who has been de-googling for awhile cannot step away from YT and PeerTube just isn't the same thing at all.


consider though that if peertube just emulates the practices of existing players the result will likely be exactly the same

the high thresholds and lack of mainstream polish of most open source projects is a true fact and has to do with the economic / business models (or non-models) so its not an easy challenge to solve

somehow being able to engage creative people to contribute (whether it is UI design, visual art, engaging text etc) will be critical for FOSS to have bigger impact


I was tempted to agree but considering they managed to make such a nice and large piece of software without marketing.. I'd keep the marketing low and let the energy flow organically.


Have you noticed YouTube heavily throttling unofficial clients lately?

Maybe we really need a video delivery service that doesn't break every other day to please beancounters at Google.


If you mean youtube-dl then you might be better of migrating to yt-dlp. It doesn't look like youtube-dl is going to resolve those throttling issues any time soon.


Nope, Google is actively targeting different tools, yt-dlp included.

It's a game of whack-a-mole right now.


Yes, if you don't provide a correct token that is calculated in heavily obfuscated javascript, the CDN will throttle you to kilobyte/s levels, basically making it impossible to watch anything real-time.

The common way to deal with this seems to be to "emulate" the function calls on the input data, which seem to be randomized.


Yes, I have a browser plugin to open youtube links in mpv, but it's been unusable recently with constant buffering. I dont even fetch them at best quality.


Even the official client has been buffering heavily on iOS for the last two days. No idea what's going on, but I'm getting RealPlayer flashbacks.


Oh goodie, I am not the only one. I ran into this issue last night and YouTube app in my iPad is buffering heavily and stuck in 480p. My usual solution for this is to restart the iPad. Yup, have to restart the iPad because swipe-to-close didn't help.


pip install yt-dlp

In ~/.config/mpv/mpv.conf, add the following line, preferably near the top:

script-opts-add=ytdl_hook-ytdl_path=yt-dlp

That should hopefully solve your issue.


You don't need to update the config if you have the latest version of mpv (0.34) released one month ago.


Are you using yt-dlp with mpv?


An alternative to centralized video sharing is definitely needed and PeerTube looks interesting.

Discoverability is one of the things that makes YouTube so compelling. Unfortunately, it doesn't look like PeerTube offers much on this front at the moment.

There's a search app:

https://sepiasearch.org/

But there's nothing like the recommendations or related video feature of YouTube. How hard would that be to add given PeerTube's architecture?

edit: I just discovered that that GitHub page does a much better job than the landing page of explaining PeerTube. In particular, it breaks down features from the perspective of users, creators, and admins. It even leads with an introductory video (but not embedded for some reason).

https://github.com/Chocobozzz/PeerTube


While I agree that recommendation are what made YouTube so popular, and also boosts engagement a lot, they are also the most controversial feature of YouTube because they tend to converge toward the most polarizing videos, conspiracy theories, etc. They've also been gamed in a few creepy ways[1]

[1]: https://medium.com/@jamesbridle/something-is-wrong-on-the-in...


YouTube literally keeps showing me the same Playlist of videos every day, and absolutely refuses to let me get anything fresh until I behave like a good cow user and watch the videos they want me to so they can milk my eyeballs with ad views.


On the web client, it's easy to mark these videos as "Not Interested." I haven't noticed them showing up again after that.


The problem is that it's the type of content that I generally want to watch. So if I mark it as not interested I'm assuming it'll generalize and not give me the same kind of content. Stuck between a rock and a hard place.

The reason I would prefer not to mark as "not interested" is that there is so much content that is being produced by channels that I subscribe to that I never get recommended to me.


PeerTube is still centralized to whatever federated instance you are using. Still waiting on a true decentralized solution.


Check out the LBRY protocol and the LBRY desktop app or it's centralized frontend Odysee


LBRY desktop is also centralized btw. AFAIK there are no front ends built by totally different teams from LBRY, although there could be.


I might be inferring the wrong thing from what you're saying, but I believe you're wrong.

The same video that is uploaded on multiple instances can share peers, so when you're watching it, you get it in a decentralized fashion.


Also videos are regular webtorrents, they can be downloaded (or watched) without any support from the instance, as long as one peer is online. Less sure about video description and comments (though comments received by your instance should be displayed).


It took me a long time to find it, but this is "the" search engine for PeerTube:

https://sepiasearch.org/

This could be a reasonable entry point for a non-technical user who just wants to watch videos. If you layer in recommendations system, trending etc. and have a default "Instance" for users to join, Peertube could start to compete with the user experience of YouTube


These recommendations you're laying out to help compete with YouTube are antithetical to the whole purpose of peertube.

First, a default instance is definitely a bad idea if you're trying to build a decentralized network of tube site servers, and it increases cost on a central entity in the network (probably framasoft who would be hosting the "default" server). The whole idea is for there not to be a central server.

Recommendations and trending are architecturally untenable as well as antithetical. When you're collecting whatever criteria makes a video "good" from disparate servers across the web you introduce a ton of bandwidth and other issues. And why would anyone want some algorithm deciding for them what they feel like watching? Beyond that, something like this is hard to get right, you're more likely to get it wrong and wind up with pissed off users.


The elephant in the room is that videos take a long time to create. And thus hosting of the videos won’t easily be trusted to random server admins.

I barely trust pleroma/mastodon admins with my garbage little text posts. I wouldn’t want to put my hard work on a server with no guarantee of how long they’ll be there or if one of them will be deleted and maybe I won’t even realize it.

Yes, youtube sometimes deletes channels or videos. That isn’t perfect either. But somehow Peertube feels even worse because the entire fediverse is run by a bunch of “some guys” that are accountable to no one.


It seems to me that a federated system like PeerTube would give you more control over which admins you are beholden since you can choose where to publish. You could even setup your own host if you don't trust any of them

Also, can't you post your videos to other platforms and/or keep your own backups? If the host deletes your video, can't you just republish it on a different one?


if you can trust yourself, host your own. that's the point


That limits the userbase to only people that are savvy enough to be a server admin, though. I've actually been discouraged from running my own server by people in the Matrix Chat world. Apparently there are risks if you're not knowledgable enough.


I like the idea of PeerTube, but I have a question:

How well does the p2p offloading of viral videos work in practice?

How is the average user experience? Do you get a lot of buffering pauses, or is it seamless as if it was Youtube? Assuming a medium-sized server hosting the content.


Actually, the experience is really fluid for most people. There can be some edgecases where we have users reporting huge bufferings but that's rare and hard to diagnostic since often bound to the client.

Before the V3, we tried the live with more than 150people watching at the same time with some friends. 1/5 of the bandwidth was served by the server, P2P handled the rest.

Since then, things continued to improve and the goal of Framasoft (from which I am external, so ask directly to Framasoft if you have specific question) is to provide the most stable experience with PeerTube.

You can find out our feedback (in french, though), on Framasoft forums: https://framacolibri.org/t/fonctionnalite-live-retour-dutili...


My experience to date has been the buffering I experience while watching a video kills the experience.


The most concurrent views I've seen on my server was about forty one, including myself. It was seamless, however the video was only around 280 megabytes and there was only one quality setting for it. I don't imagine it being the case for HD videos where users are viewing in disparate qualities.


I get excited whenever I see a new federated alternative of existing big players, like PeerTube ( vs Youtube ) or Mastadon ( vs Twitter ) or Diaspora ( vs facebook ) but unfortunately I never get around to using them considering longevity of a single/multiple nodes in the system. I know it does not make sense but fear or losing/migrating preferences/data has kept me from using most of these services


What data is there to lose? I think it's best to just jump in and try, if you want to.


How is this similar or different from LBRY / Odysee?


I don't think there's a blockchain or currency involved with Peertube. It's more like Mastodon for videos. In fact, you can follow Peertube accounts and comment on Peertube videos using your Mastodon account.


There's no blockchain involved or cryptocurrency, you do not need to pay to make a transaction, votes are one-user-one-vote rather than weighted by ownership of the currency, it is easily capable of federating with non-Peertube services due to Activitypub.

LBRY uses IPFS to achieve P2P data transfer, while Peertube uses webtorrent or HLS, with a preference towards the latter. Peertube also seems easier to set up, at least to me.


The IPFS on LBRY also means that the videos themselves don't have a "home" on one particular server. It's more like torrenting files.

Correct me if I'm wrong, but I assume that if you take down a server holding a peertube video, it's not as though clients who have seen the video (or even federated servers?) would continue serving that video. The P2P aspect is merely a bandwidth saving scheme when clients are playing the video?


Another server has to set up a video redundancy for the clip for that to happen. Otherwise, yeah, it just goes poof. Avoiding this is one of the major benefits of IPFS.


It's cool that redundancy is a built-in feature at least.


this is super nice. i have submitted patches to the project out of curiosity and they were accepted. A nice community on github. I would suggest people here to try using the website , tilvids.com is a perfect example. i was doing its beta testing of the livestreaming and i found 30 seconds delay between screen recording and viewing on a client. That was some random russian server so the latency was bad anyways because of location but still.

Tldr; tilvids.com


If this is the current pinnacle of the decentralised Web, I don't expect to see a YT competitor in my lifetime.


Please show your solution so we can follow it verbatim


I don’t think it’s the pinnacle, no. I think the fediverse has been a stopgap. Not trying to minimize the hard work of these developers, but the user experience just doesn’t make sense.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: