Hacker News new | past | comments | ask | show | jobs | submit login

Hello, i'm not a contributor to the matrix ecosystem but i'm happy to see discussion of the vocabulary used and MSC1767. Is there a working group talking about vocabulary or interop with other federated networks (ActivityPub/XMPP)? I'm a contributor to joinjabber.org which is not a software project but a community-oriented one, and we'd like to showcase and help improve interop with Matrix, but for now due to bugs in bifrost a bot relaying messages is the best we can do.

Specifically about hummingbard and/or MSC1767, have you considered borrowing/extending ActivityStreams 2.0 vocabulary? It's based on JSON-LD (proper schemas make it easier to test implementation) and could be extended to support matrix Actors/Services.

There's already a bunch of really cool stuff being developed on ActivityPub (lemmy/funkwhale/peertube/mobilizon) and Jabber/XMPP (movim/libervia). It would be great to achieve interoperability between all those.




Arathorn seems to always be on HN but chose to leave this question unanswered. I wonder why?

I'll hazard a guess: New Vector LLC sees it is in their financial interest that they own the only functional implementations of the Matrix protocol and what it can interact with, so there's no interest in federation with the wider Fediverse and other protocols.

Now Arathorn, come tell me why I'm wrong


Please don't post flamebait to HN or cross into personal attack. You may not feel you owe the other person better—but you owe the community better if you're participating here.

https://news.ycombinator.com/newsguidelines.html


sigh

The reason I skipped over this was because the author seemed to be pushing an agenda elsewhere in the thread ("Matrix is like XMPP but with a heavier server": https://news.ycombinator.com/item?id=26279816, and "Matrix has one reference client/server with really good features but hardware requirements that place it out of ordinary folks' reach.": https://news.ycombinator.com/item?id=26279780) which made me wonder whether answering all their posts was going to be a good use of time. In practice, the thread hasn't exploded into a XMPP v. Matrix holy war, so it looks like my concerns were ill-founded.

To actually answer the question from the post in question:

> Is there a working group talking about vocabulary or interop with other federated networks (ActivityPub/XMPP)?

No, there isn't a formal WG - instead there are a set of informal conversations between the 3 projects. I lurk on xsf@muc.xmpp.org and standards@xmpp.org in case interop discussions come up; I also hang around various cross-project rooms where Gargron and Christopher Allen Lemmer and others are active on the AP side. As a result, cross-project initiatives sometimes pop up - for instance Matrix supported Mastodon in adding E2EE at https://github.com/tootsuite/mastodon/pull/13820, and we've also tried to work with XMPP on using Olm for E2EE as an alternative to libsignalprotocol in OMEMO.

In terms of content vocabulary specifically: we researched JSON-LD a bunch when working on MSC1767 (back when it was MSC1225) - the notes predate us using GFM for MSCs and are archived at https://docs.google.com/document/d/1m4VTRqclB3JEMZBjbr4t5cvI...

In practice, I'm not convinced there's a huge amount of value of standardising on the same vocabulary though - whether that's MIME, JSON-LD, RDF-flavoured JSON-LD, IPLD, rich XMPP stanzas, etc - as long as there's a way to map between them. Given you need to map the protocol layer in general when bridging protocols, why not map the content format too? That said, if folks have solved the problem elsewhere satisfactorily, then we'd have no problem with piggybacking on it on Matrix (just as others have piggybacked on Matrix's E2EE).

In terms of the parent's comment:

> I'll hazard a guess: New Vector LLC sees it is in their financial interest that they own the only functional implementations of the Matrix protocol and what it can interact with, so there's no interest in federation with the wider Fediverse and other protocols.

No, my priority is trying to make the wider Matrix ecosystem as diverse and healthy as possible, including bridging nicely other protocols into Matrix. The more different projects building on Matrix (e.g. hummingbard) the more potential in the network and protocol for everyone involved. A rising tide lifts all ships, as it were. My dream for Matrix is to become the realtime communication layer of the open Web, and for there to be as many different parties building on it as do the open Web today. If we pull that off, then Element (formerly known as New Vector) would hopefully benefit from it as much as everyone else - much as 3Com benefited from creating Ethernet and bootstrapping that industry alongside DEC, Intel, Xerox and others.

But it would be UTTERLY STUPID for Element to privilege itself over other contributors to Matrix, and would effectively self-sabotage the open ecosystem, thus killing the main thing which could make Element successful in the long term. Hence the huge amount of effort and $ that we put into setting up The Matrix.org Foundation (https://matrix.org/foundation) as a neutral independent guardian of the protocol, and protect it from anyone commercially building on Matrix... including Element/New Vector.

Sorry you think we suck though :)


Sorry that you have to deal with this on literally every HN thread. I used Matrix every day, along with a number of friends and family. It has been excellent and added a lot of value to our ability to communicate. I also use the same identity to participate in the openSUSE room, PinePhone room, and others.

I have some grievances with Matrix. I've aired them in other threads, and on the Matrix.org server. Some have been addressed, others have not. I really hope I was constructive in my criticism.

Frankly, some members of the community here act like absolute children every time Matrix is discussed, and hold it to impossibly high standards (like complaining it isn't as usable or polished as privately owned VC competitors, or older standards that have been around for 5x as long and can't do half of what Matrix can). It's disappointing and frustrating because I'm afraid that you will be less willing to engage here, and upset that it detracts from actual critical conversation of the protocol.

Anyways, thank you for what you do and continuing to engage here.


Can you point to a high-level description of Matrix which makes clear where the higher server requirements come from? Obviously a lot of it is not yet optimal code, but even so what are the expensive parts in scaling up?


Synapse is relatively RAM heavy because it leans a lot on in-memory caches to speed up DB access, as the DB schema isn't particularly intelligent and the fastest way to get it up and running was to slap caches everywhere. In the past, it could also spike RAM (which python then tends not to release back to the OS) when doing complicated merge operations of complex rooms, but this has largely been solved in the last few releases. The next generation of servers (Dendrite, Conduit, Construct) use way less RAM, having built without the time constraints (and now maintenance constraints) that impact Synapse.

Separately, it's true that Matrix intrinsically uses more resources than XMPP, because Matrix's architecture is one of replicating and persisting conversation history and state between servers, rather than passing messages. And this simply involves much more data and processing than just relaying messages between sets of users.


Thanks for your hard work!


> the thread hasn't exploded into a XMPP v. Matrix holy war

Thanks for replying, that is definitely not my intention. My main concerns are interop, privacy and usability. I have technical/political criticism on both sides (XMPP/matrix) but as an end-user/sysadmin i don't see much difference in spirit/protocol between XMPP and matrix (except for decentralized rooms).

I'm currently using Jabber/XMPP for technical reasons. My two blocking issues with matrix are client side with element being really latency-sensitive (+ > 9MB loading), and server-side with synapse taking resources so the non-profits i know (eg. tildeverse.org/chatons.org/libreho.st) only provide Jabber/XMPP service. But like i said elsewhere in the thread i'll give dendrite a go.

My agenda is not hidden and is as simple as empowering users to federate across decentralized networks/protocols so we can stop arguing which solution is the best and stand up together against the corporate silos. This was stated in the goals of the joinjabber.org collective we founded a few weeks back: https://joinjabber.org/collective/about/goals/

I can assure you i have bitterness on both sides, with XSF mostly concerned with business use-cases (not free/non-profit ecosystem), and with matrix investing considerable resources into reinventing the same wheel as XMPP did two decades ago (a universal federated, bridgeable messenger). Despite everything, my agenda is as simple as "interoperability".

> as long as there's a way to map between them

I agree on paper. But in practice very few projects actually do that (only Hubzilla/Bridgy to my knowledge) which led to my question of potentially reusing an existing standard instead of coming up with an entirely new vocabulary. In my personal view, coming up with new vocabulary is also a valid approach if interop is not an afterthought.

> if folks have solved the problem elsewhere satisfactorily (...) to become the realtime communication layer of the open Web

I believe ActivityPub/ActivityStreams has solved many problems, and is already the most advanced/standardized federated social protocol on the open web (it's a W3C recommendation) despite not being especially real-time oriented. There is room for interpretation/improvement in the specs but "user profiles, posts, communities, sharing" (as in hummingbard) is definitely covered by major implementations in the fediverse even if "communities" use-case was not explicitly mentioned in the initial AP spec.

Still i made the effort to follow your link to Google Docs (seriously why Google? i made an exception for you but it's fucked up) and i better understand your concerns about reusing/mapping ActivityStreams to matrix vocabulary. So while i can't say i encourage you to reinvent yet another social federation vocabulary, having a strict mapping like you mentioned in the doc makes sense... but it should really be priority n°1 otherwise we're just splitting the ecosystem even more (see XKCD about standards).

Interoperability for social networks is really cool. People enjoy being able to comment on a Peertube video directly from their Mastodon client/account. I hope we can have matrix spaces (and hummingbard and others) part of this federated future as well. I don't believe standardizing a gateway/bridging mechanism is enough if interoperability is an afterthought (in the rest of the standards) because that's precisely the path that Jabber/XMPP took ~15y ago and it's failed dramatically.

Take care and keep in mind despite our respective opinions/criticisms, we all care for empowering and protecting users. Our enemies are corporations, governments and other kinds of oppressors, not other federated protocols.




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

Search: