Her sidder jeg, med mit hjerte brudt // Prøvede at skide, men slog kun en prut
Is renaming the instance domain without reinstalling Lemmy related to changing the WebFinger query? It’s the trick some instances use to have a different instance domain from their username domain, like @user@domain.com while the instance is mastodon.domain.com.
There should be a patch for it that hides the “recommended” feed in the homepage. I’m not certain because I never use Youtube with an account or the official website/app, so I don’t get targeted recommendations.
That’s why I suggested Revanced with “disable recommendations” patches. It’s still Youtube and there is no new platform to learn.
I watch a ton of videos there, literally hours every single day and basically all my recommendations are about stuff I’m interested in.
The algorithm’s goal is to get you addicted to Youtube. It has already succeeded. For the rest of us who watch one video a day, if at all, it employs more heavy-handed strategies.
I think it’s sad how so many of the comments are sharing strategies about how to game the Youtube algorithm, instead of suggesting ways to avoid interacting with the algorithm at all, and learning to curate content on your own.
The algorithm doesn’t actually care that it’s promoting right-wing or crazy conspiracy content, it promotes whatever that keeps people’s eyeballs on Youtube. The fact is that this will always be the most enraging content. Using “not interested” and “block this channel” buttons doesn’t make the algorithm stop trying to advertise this content, you’re teaching it to improve its strategy to manipulate you!
The long-term strategy is to get people away from engagement algorithms. Introduce OP’s mother to a patched Youtube client that blocks ads and algorithmic feeds (Revanced has this). “Youtube with no ads!” is an easy way to convince non-technical people. Help her subscribe to safe channels and monitor what she watches.
Using Piped/Invidious/NewPipe/insert your preferred alternative frontend or patched client here (Youtube legal threats are empty, these are still operational) helps even more to show you only the content you have opted in to.
Lemmy: Oldest federated link aggregator, better documentation compared to Kbin, easy to self-deploy, less resource consumption, provides the most similar experience to Reddit
Kbin: Poorer documentation, no API access yet, harder to self-deploy, terminology and UI differences from Reddit can turn people off (I really don’t like “magazine” for a community)
Tildes: Centralized, invite-only and elitist. Not comparable to Lemmy and Kbin
Why do you have to use NGINX? Caddy does the proxying to the Lemmy containers for you. That docker-compose.yml file is my entire deployment, there is no hidden NGINX container or config file that needs to be added. Just remove your broken Lemmy deployment with docker compose down and delete the containers, then docker compose up my docker-compose.yml (after you edit the postgres variables) with config.hjson in the same folder.
Oh shit, I forgot that your Caddy would be running on a bridge network by default because mine is on the host network where all ports are already exposed to it! (It’s generally a bad idea to use the host network, so don’t do this if you’re only using Caddy with containers on the same network) I edited the Gist to expose 80 and 443 for HTTP/S on that container, the updated file uses the same Github link. Really sorry about that!
Ugh, that one is a problem with indentation because pasting the config into a Lemmy comment destroys the formatting. I uploaded it on Github to preserve the correct indentation.
Yeah, the config file on the documentation sucks. I had to poke through several discussions on /c/selfhosting to find a config that wasn’t the extremely minimal one linked in the documentation. Your config.hjson is fine from what I can tell, although I’m not sure why you censored the hostname there as it’s supposed to be lemmy.emphisia.nl and not anything confidential.
Honestly, I don’t have enough understanding of NGINX to debug its config, so I’ll just share my docker-compose.yml for leddit.danmark.party which worked correctly and federated out of the box, with a few adjustments to match your deployment. Note that you’ll have to tear down your existing deployment if you want to use this docker-compose.yml because they use the same ports.
version: "3.9"
x-logging:
&default-logging
options:
max-size: '10m'
driver: json-file
services:
caddy:
image: caddy:2
volumes:
- ./volumes/caddy:/data
- ./volumes/caddy:/config
# See Caddy's documentation for customizing this line
# https://caddyserver.com/docs/quick-starts/reverse-proxy
command:
- /bin/sh
- -c
- |
cat <<EOF > /etc/caddy/Caddyfile && caddy run --config /etc/caddy/Caddyfile
{
debug
}
(common) {
encode gzip
header {
-Server
Strict-Transport-Security "max-age=31536000; include-subdomains;"
X-XSS-Protection "1; mode=block"
X-Frame-Options "DENY"
X-Content-Type-Options nosniff
Referrer-Policy no-referrer-when-downgrade
X-Robots-Tag "none"
}
}
# Lemmy instance
lemmy.emphisia.nl {
log
import common
reverse_proxy http://lemmy-ui:1234 # lemmy-ui
@lemmy {
path /api/*
path /pictrs/*
path /feeds/*
path /nodeinfo/*
path /.well-known/*
}
@lemmy-hdr {
header Accept application/*
}
handle @lemmy {
reverse_proxy http://lemmy:8085 # lemmy
}
handle @lemmy-hdr {
reverse_proxy http://lemmy:8085
}
@lemmy-post {
method POST
}
handle @lemmy-post {
reverse_proxy http://lemmy:8085
}
}
EOF
lemmy:
image: dessalines/lemmy:0.18.1-rc.9
ports:
- 8085:8536
volumes:
- ./lemmy.hjson:/config/config.hjson
depends_on:
- postgres
- pictrs
restart: always
logging: *default-logging
lemmy-ui:
image: dessalines/lemmy-ui:0.18.1-rc.9
ports:
- 1234:1234
environment:
- LEMMY_UI_LEMMY_INTERNAL_HOST=lemmy:8085
- LEMMY_UI_LEMMY_EXTERNAL_HOST=localhost:1236
depends_on:
- lemmy
volumes:
- ./volumes/lemmy-ui/extra_themes:/app/extra_themes
restart: always
logging: *default-logging
postgres:
image: postgres:15-alpine
ports:
- 5432:5432
environment:
- POSTGRES_USER=MyPostgresUser
- POSTGRES_DB=MyPostgresDb
- POSTGRES_PASSWORD=MyPostgresPassword
volumes:
- ./volumes/postgres:/var/lib/postgresql/data
restart: always
logging: *default-logging
pictrs:
image: asonix/pictrs:0.4.0-rc.7
user: 991:991
hostname: pictrs
environment:
- PICTRS__MEDIA__VIDEO_CODEC=vp9
- PICTRS__MEDIA__GIF__MAX_WIDTH=256
- PICTRS__MEDIA__GIF__MAX_HEIGHT=256
- PICTRS__MEDIA__GIF__MAX_AREA=65536
- PICTRS__MEDIA__GIF__MAX_FRAME_COUNT=400
volumes:
- ./volumes/pictrs:/mnt
restart: always
logging: *default-logging
postfix:
image: mwader/postfix-relay
environment:
- POSTFIX_myhostname=lemmy.emphisia.nl
restart: "always"
logging: *default-logging
I don’t use NGINX as my proxy server, but it’s a bit strange that you would need two configs for this while mine runs perfectly with one config and two open ports (:8536 for Lemmy-BE and :1234 for Lemmy-UI). And why are you using different versions of Lemmy-BE (18.1-rc9) and Lemmy-UI (18.1-rc4)?
If you are using the default docker-compose.yml on the Lemmy repo, that part of the NGINX config uses https:// + the name of the Docker containers. And you always give NGINX the external port (the number on the right side of the colon defined in ports:, like 1234 in 1234:5678). The port on the left is only known to the container the port is defined for.
If it’s still broken after you correct the NGINX config, what are your docker-compose.yml and config.hjson like? There’s several versions of them floating around and you might have combined incompatible versions with each other.
“Suddenly”? This has been happening for a long time. If you click on outbound links from built-in Windows apps, they used to always open in Edge unless you used a tool named EdgeDeflector to redirect them to your preferred browser. In 2021, they killed EdgeDeflector by making it impossible to redirect links with the microsoft-edge:// protocol baked in, even if you go deep into the registry settings to change this. They will eventually do this to Outlook and Teams too and get away with it, just like they got away with restricting EdgeDeflector.
Instances can be scaled across several machines. Here they’re using multiple containers as their resource usage isn’t high enough to require multiple machines yet, but it proves that it can be done.
Yes, it started from this terminology change at Twitter in 2020. They’re the reason that version control systems call the primary branch ‘main’ instead of ‘master’ by default, because ‘master’ comes from the master/slave terminology that is used in electronics hardware design.
There’s a comment here saying that master/slave in hardware design is being replaced by primary/secondary because of the software trend, which I think is stupid. Master/slave works much better in that context because the master device controls the slave device. Primary/secondary implies that the slave device is a fallback of the master device.
Pixiv does not allow direct linking to images as it uses an authentication token to fetch images, and this token is only sent if you view the image from the pixiv.net site. They don’t have a “no token required reduced size preview” version of the image that is used to generate thumbnails. What @iraldir saw the second time is probably a cached version of the image as it always returns 403 when I request i.pximg.net directly.
I just look to the microblogging side of the network (which has about 10 million total users) as a case study.
The ideal situation? More nodes are added to the network to spread the load and control away from a few very large and very expensive instances. The realistic situation? Some instances manage to secure external funding (such as mastodon.social) and grow extremely large at the expense of smaller instances that shut down from a lack of users and funding. Decentralized protocols like the fediverse and email are not immune to centralization thanks to lazy users who join the biggest instance. My pessimistic outlook is that the Fediverse will eventually become like email, with a few very big instances and a lot of spam making it difficult for smaller instances to enter the network. Enjoy the fresh new internet feeling while it lasts and move on when the platform starts to decay.
It’s great that they’re going back to traditional, self-hosted forums instead of corporate social media for support and discussions, but damn, I don’t miss having to manage hundreds of accounts with unique logins for each forum. I understand that they want more control over forum moderation and the Fediverse’s “anyone can post there” system makes it troublesome. It would be great if there was more widespread adoption of decentralized, “one login to access everything” systems.
This instance is hosted in Germany, one of the countries with the strictest anti-piracy laws? Seems like a very risky decision (I’m aware that a lot of the good and affordable hosting providers are German).
No, it’s 100% economics. Why do you think that having “careers, lives and travel” (as if having a family is not having a life?) is more appealing to modern first worlders? Because it doesn’t impact their finances severely. Having more children in impoverished countries is a financial gain because children are free labor and lottery tickets to get the entire family out of poverty. In wealthy countries, children are only a financial loss.