Without constraining yourself to Mastodon/Misskey/Pleroma/Frendica/GoToSocial/other current social implementations and/or the current features of them (meaning quote and threading is already accounted for in the question) what would *you* want that you haven't seen?

Think of the next thing, as the above are all built (or in-progress) sets of the virtual "Lego-kit". With the possibilities out there to play with- what new thing would you come up with?

@chillicampari
Not really a new thing but I guess I’ll throw secure messaging into the mix, although I have to wonder if federated systems could accomplish it. Decentralisation and untrusted servers seem a minimum requirement to achieve the meta data resistance required for a truly secure messenger so I’m uncertain that it belongs in this discussion. (Not a coder ;) just love the theorectical and functional perspective on this issue)

@chillicampari while i like the home/local/federated feeds of mastodon, i feel like there is space for a "neighborhood" feed of servers who have found eachother to meet similar moderating standards and cultures or have similar themes (a lot of art instances are closely linked). I know I've phrased it as a mastodon thing but I'm really talking about webrings.

@chillicampari
With people actively implementing quote-and-boost I would want a pop-up with choices when tooting.
a) Do you want to allow others to quote-boost?
b) Are you sure you want no content warning - this server has a policy of .....
c) You did not add alt-meta-description-tags for your media . Are you sure?
d) #addyours
It should be small, unintrusive but I think it will be necessary.

@chillicampari federated club penguin (please not built on activitypub though, that would be cursed 😆)

@chillicampari live chat, using voice, spatially organised, where someone who speaks English can chat with people who speak foreign languages, for example Japanese people, or the French, or Americans, and it live-translates. It might be an excellent language-learning facilitator (or the opposite, nobody need ever learn any other language than their own)

@chillicampari Linking accounts across fediverse implementations so that you could, for example, post from a single account and decide which server it published to.

@chillicampari I'd love to see a federated alternative to Thingiverse and Printables (I did build this for MediaGoblin long ago, but I don't think the feature caught on).

I'd also love to see games use ActivityPub in interesting or useful ways. There's some obvious uses like match making or posting scores, but I think there's room for creativity here to do much more interesting things than that.

@chillicampari honestly the biggest feature i want is being able to have some sort of shared/unified timeline and notification system, for multiple profiles, or possibly better limiting who can see what (but also somehow self service)

for example i want to have like a standard masto profile for text stuff, and then a photos profile or stream, and allow people to just follow the photos stream, but be able to manage them from a single pane

@chillicampari This: https://deadsuperhero.com/2022/02/towards-a-greater-federated-architecture/

Specifically, better identity management so that users can have one account to access many different frontends, and much more robust access controls on posts for their privacy.

@CallMeWeezy @chillicampari Totally agree -- toot.cat admins mentioned this in their "Federating with the trouble" presentation back in 2017 and it's really bizarre that it hasn't happened yet. Akkoma has a concept called "bubble" that seems like it's somewhat along these lines, but I don't know the details.

@chillicampari
Ah I may have misunderstood the "whithout constraining" part.

On a meta(!) layer I am waiting for Rifkin's thesis to manifest, with more meshes, decentralized networking, energy production and stuff. For that to work, we would need a new kind of TCP/IP I guess, not even IPV6 stuff. But I am not the expert for that. However, "distributed anything" with federated authentication is what I think we need - but seriously no #blockchain voodoo.

@markcharnell @chillicampari That already exists, it's called Matrix and it's quite successful and has been around for a little while already.

@lrhodes @chillicampari strong agree. being able to manage multiple accounts (or even multiple views of the same account) would be so nice

@jdp23 @CallMeWeezy @chillicampari Bubble allows you to add certain instances that you consider "related" and that has its own timeline.

@jay @chillicampari @CallMeWeezy thank for the response -- and that sounds great! are bubbles added at the instance level or by an individual?

@jdp23 @chillicampari @CallMeWeezy It's at an instance level. (I run my own Akkoma single-user instance and was about to explore this very thing myself shortly.)

@jay a question on Akkoma: what resources are you finding it requires?

@jdp23 Akkoma is a Pleroma fork, so you can look around for what people say for resource usage for Pleroma also.

For my single user instance with something like 20 relays enabled on a server with lots of other things running (Nextcloud, Owncast, Navidrome, etc...) it's almost never noticeable. It'll pop up at the top of htop but with a CPU usage of 2.9% or something like that on my colocated server. In terms of CPU/RAM, it is far far lighter than Mastodon. Light enough that you can comfortably run it on a Raspberry Pi if you wanted to without issue.

While Mastodon is far heavier for CPU and RAM usage, the real resource issue for it is disk space. With about half the number of relays and with Mastodon set to expire and remove cached statuses after 7 days, I was still running between 50-100GB in disk space. My Akkoma instance with 10 more relays enabled: 4GB.

This is because Pleroma and Akkoma do not cache profile and image graphics locally at all and only pull them down when needed, as far as I am aware. Seems to do a good job of it too as it is rare that I see any sort of missing images from posts.

Hope this helps.

@jay yes, very much so! I've seen descriptions elsewhere of how much Mastodon pulls down and then doesn't delete so I can certainly imagine there's room for improvement.

How much ram do you have?

@jdp23 My server has 32GB of RAM and it looks like Akkoma's got several processes but totaling less than 1GB, I think. The numbers I see are far far less than 1GB but I don't have the time to check every process right now.

@jay cool thanks again! @fahim was talking about memory usage at
https://a.farook.org/objects/be31ada4-7512-4990-b94f-7248ffa72c2f so I was curious what your experiences are

@jdp23 @fahim I get a 404 on that link. Yeah, it's REALLY light.

@jay @fahim oh huh, it must be a bug -- I just verified that's the link I get when I do a "copy link" on Fahim's post.

Well okay then, here's my reply -- hopefully you can see his parent post.
https://indieweb.social/@jdp23/109598559402284331

@jay Without sounding as if I wanted to pick a fight, I did want to say that "light" is all in your context 🙂 If you have 32GB of RAM on your server, then you'll be totally fine. But if you are using 1GB RAM VPS, you'll struggle which is where I ended up ... It's not Akkoma's fault in that Akkoma itself does not take a lot of RAM. But in my case, I believe the struggles were because of Postgres, which did seem to require a fair amount of RAM.

At the moment, I have Akkoma running on a 1GB RAM VPS without issues after enabling cache. Postgres takes around 360+ MB of RAM and the overall RAM usage (when I checked yesterday) was around 85% of the total. And I'm not running anything else on that server besides Akkoma, Postgres, Nginx and anything else that comes pre-installed on a basic server.

But somebody else told me (based on their experience) that they didn't even think you could run Akkoma without 2GB of RAM. So yeah, it probably is light compared to Mastodon, but it's all relative 🙂 They all run fine locally on my MacBook which does have 32GB of RAM, but when I move to a VPS, that's when I start having issues ...

@jdp23

@fahim @jdp23 Postgres for my Akkoma instance is running 180MB of RAM according to pmap.

@jay Interesting. Did you do any tweaks to your Postgres install? Or is it running the default config? I did all the optimizations that Akkoma docs suggested but kept the connection count at 20 instead of lowering it. So wondering if there's anything else I could do to get the memory usage you're seeing ... @jdp23

@snott @jay thanks! Did you do any tweaks to your Postgres install? @fahim

@snott @fahim @jay Thanks much!

@fahim @jdp23 No tweaks at all. Using about 20 relays. Not using emoji steal.

@snott @fahim @jdp23 No changes at all from default on my side.

@snott Thank you for this — it's helpful to compare with other installs to try to figure out where things might be tweaked 🙂 For comparison purposes, I've attached a screenshot from btop for where my server is sitting.

It looks as if you are running Postgres 13 (while mine is on 12). Akkoma docs mentioned that there was an issue with older Postgres but they mentioned a fix and I implemented that. Didn't seem to help much there ... But I'm tempted to try upgrading Postgres to see if perhaps that helps?

If you don't mind me asking, what is the OS you're running? My box is on Ubuntu 20.04 I believe ...

@jdp23 @jay
btop screengrab from 1GB VPS ru…

@jay Cool. Thanks 🙂 Just trying to figure out why my Postgres has like double the RAM usage of yours. Which version of Postgres are you running?

I don't have any relays at all I believe. And didn't even know about emoji steal 😛

@jdp23

@snott Well, obviously you're doing great since it's been running for over a month did you say? 🙂 I'm tempted to spin up a Debian 11 VPS just to see if Postgres and/or RAM usage is less on a Debian 11 instance ... But I'm not sure I have the will to do so at this point since I've installed Akkoma like 20 times now over the last 3 days 😛

@jdp23 @jay

@fahim @jdp23

Here's my info from my docker compose:

akkoma_db:
image: postgres:12-alpine
hostname: akkoma_db
container_name: akkoma_db
healthcheck:
test: ["CMD", "pg_isready", "-U", "pleroma"]
restart: always
ports:
- "127.0.0.1:55432:5432"
networks:
- akkoma
env_file:
- .env.akkoma_db
volumes:
- akkoma_db:/var/lib/postgresql/data

@snott @fahim @jdp23 Thanks for bringing up btop, by the way. I wasn't aware of its existence before. Kind of nice even if the spacing doesn't quite look right on SSH.

@snott @jdp23 @jay I think I just might try Debian 11 to see if I can get better memory performance. I feel better if I wasn't so close to edge in terms of memory usage on this box 🙂 Just gotta find the time I guess .,..

@jay @jdp23 Thanks! I think that's Postgres 12 .. the same version as mine. So it's not the Postgres version that affects the lower RAM ... At least, doesn't look like it. Now this'll keep bugging me till I figure it out 😛

I'm basically running everything without Docker at the moment since I didn't want to have the overhead of Docker on the machine, but I'm still curious as to why I have much higher Postgres RAM usage ... Ah, well ...

@snott @jdp23 @jay Good luck with that 🙂 It's always fun to try out these things ... when you have the time.

I'm currently looking at migrating both my servers (one 2GB and the Akkoma one at 1GB) over to a single 4GB server. I have the next week off and so that seems like the time to do it, but man, I don't know if I want the pain 😃

@snott @fahim @jdp23 Would like to hear about your opinions on write.as when you get the time.

@snott I'm fairly sure you are bad for my sanity ... and I just met you 😛 You keep saying, "Do it! Do it!" and I keep thinking, "Hey, why not?" ... 😃

@jdp23 @jay