Same but with XMPP! Much easier to host (there is an order of magnitude difference in resource consumption). Although my client can't do live location, only manually sending the current point.
Why is the resource consumption of the Matrix server implementations so high? What are they doing that hits so heavy on performance? Encryption? But isn't it e2e? State resolution? Surely the resolution algorithms can't be that bad?
Even Dendrite, the eternally-WIP-rewrite in Go, used so much RAM I couldn't even start it on my Raspi.
The reason Matrix typically consumes more resources than XMPP is because it's replicating a large DAG of state events (i.e. the key-value data associated with a room; not the room's actual messages) between all participating servers. So, in a room with 100K users, each server has to maintain at least 100K state events (every time someone joined a room, or changed their name, or left the room, or set permissions, or changed the room's name etc).
Then, whenever the state of the chatroom changes (e.g. you join it, or someone else joins it), each server has to resolve its view of the state of the room with the proposed change. We call this merge conflict resolution algorithm 'state resolution', and it's typically where the heaviness comes from, as a naive implementation loads up all the state events for the room in order to calculate and store the new state.
In exchange, you get fully decentralised access control, and an eventually consistent zero-trust distributed key-value store, with resilience to network partitions (including in P2P environments) - unlike (say) vanilla XMPP which applies access control and KV on a single given server per-conversation.
Now, the key thing is that *there is nothing in the protocol which makes this intrinsically heavy to implement*. The only reason that servers can be heavyweight today is simply because more time has been invested in chasing down more pressing issues like encryption issues, investing in client SDKs, speeding up sync responses, switching to OIDC for auth, etc.
The problem is made worse by the fact that Synapse (the original Python server) has historically been particularly clunky, skewing perception. Over the last few years this has improved significantly, although it could still be way better - at the Matrix Conference last month (https://youtu.be/kdEDxI2MCjA?list=PLFDUAtPqCX98&t=329) we showed off a fork which had improved CPU 10x by shifting the hot path entirely to Rust from Python. (It turns out that over 2026, we wrote as much Rust for Synapse as we did Python, ignoring tests). Further improvements would be to memoize the state res and calculate it incrementally rather than starting from scratch every time. Similarly, there are 100x improvements to be had on storage usage - e.g. by storing state temporally (c.f. my https://docs.google.com/presentation/d/1pKtLl4vCV3-8xz8crvxW... spike from last year).
So, why haven't these improvements landed yet? In practice, the team at Element who builds Synapse have had to focus on customer-driven work instead in order to keep the lights on, and frustratingly performance is lower down the todo list than other feature-work. On the plus side, over the last year or so Element has got into a much healthier state - hence us being able to show off the 10x improvements in Synapse last month (although they still need to land).
Meanwhile, elsewhere in Matrix there are other servers which are more actively focused on being lightweight right now (e.g. continuwuity), but don't have the maturity of Synapse (or the commercially supported distribution that Element provides). And even then, there's still a lot of headroom for speeding up and shrinking things - it should be very very doable to run Matrix on a tiny host like an RPi.
It's not a perfect analogy but I think of it as being similar to the evolution of DVCS systems. There's nothing intrinsically heavy or hard about replicating DAGs of source code between hosts, but Monotone was pretty slow and clunky while Git showed that a smarter implementation could be more than performant enough. The race is on to build the equivalent for Matrix.
This design is the problem. With other protocols, a room is associated with a server, which is authoritative for that room, and no blockchain state merging needs to be involved. Messages may short-circuit at leaf servers, but only the central hub controls room permissions, which is the problem that makes it hard.
My concern is also storage. It just keeps growing and I don't know to what exact extent it can be cleaned up. My VPS only has 20 gigs of storage, which is never an issue with XMPP (because the messages are ephemeral by default and media has an expiration date). Even text can add up over a really long time in active chats. Synapse seems to have better cleanup options (seen fewer retention options in, say, Continuwuity), but overall would like more insight into the possibilities.
You should try alternative homeserver implementations of Matrix! From private benchmarks, https://continuwuity.org/ can serve over 1,000 simultaneous clients on 1gb of RAM.
I used to use Conduit, but that broke some rooms with Synapse users (apparently Synapse deviates from the defined standard somehow). Besides that - yeah, that is a much more appealing option! Although I still don't know how well it would clean up storage (as I have a very constrained disk space). Seen some media options, but not sure how you clean up messages.
Also so that any electronics don't get fried. Wi-fi board borking itself would be good, but I'd be worried about it affecting something else down the line.
Exactly. That's part of what I meant by "none the wiser." The other being that the xmitter could in theory detect a missing antenna from the impedance mismatch and try to increase its power (assuming it knows how to do so without frying itself).
"(assuming it knows how to do so without frying itself)."
Unlikely, as Wi-Fi is flea power (not high enough to do damage). Moreover, Wi-Fi chips have an inbuilt power limit so as to comply with spectrum management regulations.
Anyway, why wouldn't you want to fry the Wi-Fi chip if possible, that'd solve everything? If I had that problem I'd simply stuff the chip or short out its RF output (bridge solder across its antenna pins).
There are multiple ways of fucking the chip, likely the easiest is shorting B+ to the high side of its antenna pins (that'll likely do the trick). Another method is to hold the tip of a hot soldiering iron on the center of the chip until it fails.
Of course, you shouldn't do this unless you know what you are doing. The Wi-Fi chip may also control Bluetooth so if you still want that risky comms pathway then consult the chip's datasheet first.
I wonder how likely it is though? The consumer access points don't come without passwords by default. Where I am, having an open access point without identifying the users is illegal, so it's unlikely a regular person would do that, and if some cafe does - it'd just have a captive portal asking for a phone number. And even if the business owner doesn't want to bother with that identification - there would 100% be a password on such a hotspot.
Fully anonymous open WiFi remains legal and common in many countries, including the US.
Even many guest or public WiFi captive portals in the US only seek to enforce agreement to a terms of service and/or privacy policy, not necessarily gathering identifying info and not locking behind a password. (Yes, there are also those which do require a password or identifying info, but that’s not due to a legal requirement.)
It’s relatively rare nowadays that I see a guest or public WiFi network with neither a captive portal nor a password, but again, those do still sometimes occur and are fully legal in the US. Organizations (and their insurers) do tend to like the protections of legalese and/or identifying users, but US law doesn’t generally force them into it in this context.
Much better for any westerner. The Chinese will not divert any flights, because of what I said online, nor will they refuse entry into the US or partner country, or search my company devices.
I don't care about China, because first I will hardly travel there, nor do I expect any problems if so. Unlike GB, Turkey, Israel or the US.
Last time I looked, it didn't even allow installing apps from outside AppGallery at all. Also, I don't think the Huawei services are any less intrusive than privileged Google Play.
First - a lot of things are just not on there. And second - not allowing alternative sources gets really bad when said store is asked to censor apps, like, say, proxy clients or alternative social media apps (Newpipe, etc.).
Books are also the one thing I much prefer to have physically! I can usually get fiction in my language (mostly in secondhand bookstores now), but I still have to get most things digitally. Most books I read are in English and my country isn't English-speaking, so getting them would cost a fortune. And sometimes out-of-print books even in my language are also hard to find due to scarcity. So pdfs and epubs it is.
Not only are CDs the backup, but in some genres like classical music, the liner notes have some exclusive information that never made it to the web. I rarely take a CD down from the shelves where they now serve as mere room decor, but when I do, I often learn something new.
Even though I've ripped my 1000+ CDs, I tend to not let company dig through them, as I consider them my pristine backups!
I also buy a lot of digital music on bandcamp.com, and like OP, host using navidrome.
But you are right, when company comes over letting people DJ is fun. Instead of using my actual CDs, I set up a system with RFID cards. You can buy hundreds of RFID cards for super cheap. I print out the cover art of each of them, and the RFID ID corresponds to an album. A simple RFID reader hooked up to a pi, which reads the ID and plays the corresponding album with mpd on my stereo, makes for a cheap, fun way to allow people to physically explore your collection.
Really like the idea of RFID cards - I was brainstorming something like that years ago for the kid, but never chased it down.
(Un??)fortunately, I am usually the one slinging CDs for people to listen to, people just like being able to offload that and listen to curated music, vs. "throw on spotify whatever" and having it just _play_.
People say this and when I look it up it's supposedly true, but I grew up in the heyday of cds and cdrws and have probably burnt thousands of them including specifically expensive plextors or whatever brand wasn't bulk trash back in the day and constantly had issues with scratches, sunlight, whatever damaging them to uselessness. I was the kid selling mp3 discs and pc games when nobody could afford a cdrw burner.
There is a reason people can tell you how exactly how many times they've bought greenday dookie, offspring smash, 2pac all eyez on me and other 90s gold. I think this is some collective brainrot or everyone is talking about hiding their discs in the canticle of leibowitz and not their corolla.
edit: 35 year old cds is 1995, that your dad comment made it sound even older. I guess you're not the prime cd burner in the house, your dad was, ask him how many times he's bought those working discs albums. Or he was the guy with the giant bookcase envelope of CDs he treated like gold. I could always burn a new one.
Similar to the OP, I buy discs, rip them, then store the physical copy for recovery purposes. I keep them in a box in the basement, the vast majority are still in working order. (Including discs purchased or burned 35 years ago.)
Like any other media, frequent use and handling will degrade discs, and sunlight is the natural enemy of anything plastic. The interior of a car will also get ridiculously hot on a sunny day, I have seen vinyl literally melt. (And crayons!) Actually I've lost more discs to theft than anything else, by a large margin.
I don't even think this counts because then you'd be paying for the brittle and useless big folding display as well, when all you want is the smaller outer one. Plus the worse repairability.
To be fair, most phones have invasive Google/Apple/Chinese OEM services that work with high privileges and are invasive on their own, without apps. The share of people using degoogled OSes is small enough to assume it isn't the case. And, to be fair, if someone has a degoogled phone, you'd probably know that by now (I say as also a degoogled user XD)
Statements like this are a stretch. This was not the fact of using the OS, but specifically the fact of wiping it. The situation is atrocious enough as is, no need to make up additions.
reply