3 ms·

The difference is that there are multiple different XEPs performing similar operations, and even with the compliance suite XEPs, it can be confusing to synchron

The difference is that there are multiple different XEPs performing similar operations, and even with the compliance suite XEPs, it can be confusing to synchronise on whether different bug-free clients are speaking the same dialect. Whereas Matrix ensures that there is only ever one way of doing a given operation at any point, and those operations are backwards compatible, because there's only one spec. Obviously the client and server you're talking to needs to have actually correctly implemented the bits of the spec that you're trying to use - but there's no openpgp v. OMEMO or rival file transfer mechanisms etc. It's only a difference in governance, but it's an important one. Obviously, a single monolithic spec controlled by a committee like the matrix spec core team has its own problems (most importantly, the committee can become a bottleneck); and we still provide mechanisms to let people experiment via the equivalent of vendor prefixing, which can go horribly wrong (c.f. aggregations in Matrix) if not managed with discipline. But, I'm not trying to spread FUD about XMPP - just try to explain that the approaches taken are different, and have different tradeoffs, and are not "exactly the same".

petre 6y ago
> Whereas Matrix ensures that there is only ever one way of doing a given operation at any point Python used to claim the same thing when it was a fresh new language, as a reaction to Perl where the opposite was encouraged. Becoming mature and anchored in reality changed some of that.
arendtio 6y ago
> Whereas Matrix ensures that there is only ever one way of doing a given operation at any point I trust you that you mean what you say, but I wonder what an implementation would look like (from a user-perspective), that supports something like OpenPGP -> OTR -> Ratchet. I mean, upgrading security over time is mandatory and when I remember how it evolved for XMPP I clearly do not want an implementation that still has all the draw-backs that OTR had. So I agree that XMPP is lacking some good governance, but I don't see the rivaling XEPs necessarily as the core problem, as many of them had quite some time between the drafts. The bigger issue is that many clients don't have enough developer momentum, so that XEPs that were drafted like 5 years ago are not implemented yet. Regarding implementation bugs: I know XMPP has the compliance suite, but I feel like it doesn't catch bugs and is more of a basic testing for feature compatibility. Does Matrix have some kind of test suite, that simulates physical disconnects, package loss and the likes for testing real client and server implementations? I know that would not be an easy feat, but I wonder what the best way would be to build something like that, to improve the quality of the existing implementations.
- pseudalopex 6y ago
  OpenPGP could mean XEP-0374 which is newer than OTR. But XEP-0374 went from experimental to deferred 3 years ago. It isn't really competing with OMEMO.
  - arendtio 6y ago
    XEP-0374 is very fresh compared to other PGP-related extensions. Just to give an example, 'XEP-0027: Current Jabber OpenPGP Usage' became active in 2002. If it is competing with OMEMO or not might be a separate discussion, but both can be used to utilize E2E encryption of messages.
    - pseudalopex 6y ago
      XEP-0027 became officially obsolete in 2014. I'm giving them the benefit of the doubt they didn't mean anyone has to think about XEP-0027 in 2021. They implied they could break compatibility some day. So I think they're talking about not having 2 current standards. How they see a transition working is still a good question though.
xorcist 6y ago
> Matrix ensures that there is only ever one way of doing a given operation That's not entirely true though. Press voice call in a client and you may end up with an embedded jitsi, some WebRTC thing, or nothing at all. That's exactly how xmpp clients started to diversify. > I'm not trying to spread FUD about XMPP Yeah, the tone could certainly be improved. Every time Matrix comes up, someone mentions xmpp and immediately the flinging starts about how broken it is and how those developers let it slide into obscurity. Which there may be some truth to, but it really leaves a bitter taste. Especially since much of xmpp lives on in Matrix in bridges to other networks. Matrix really got a flying start there, and that's a good thing! Software should be shared and built on. What worries me a little is that the same problems xmpp had, with protocol verbosity and problematic integration with smartphone platforms, is bascially the same in Matrix. The protocol is at least as verbose, and battery management is at least as hard. As a successor to xmpp, one might have expected to learn more from its problems. I'm certainly not complaining however, the more software the better, and the web client is very nice.
- Arathorn 6y ago
  > That's not entirely true though. Press voice call in a client and you may end up with an embedded jitsi, some WebRTC thing, or nothing at all. That's exactly how xmpp clients started to diversify. This really isn't an example of protocol fragmentation - it's behaving precisely as expected. There is precisely one way to do a 1:1 VoIP/Video call in Matrix: https://matrix.org/docs/spec/client_server/r0.6.1#voice-over-ip https://matrix.org/docs/spec/client_server/r0.6.1#voice-over.... It hasn't actually changed since 2015 when we first added it to the spec, until recently when there's a proposal to improve it called MSC2746: https://github.com/matrix-org/matrix-doc/blob/dbkr/msc2746/proposals/2746-reliable-voip.md https://github.com/matrix-org/matrix-doc/blob/dbkr/msc2746/p.... Now, this is a proposal to patch the spec - it's a Matrix Spec Change. It retains backwards compatibility with the current behaviour, and once the MSC is approved it'll be merged into the spec and the MSC will be discarded. There is precisely one way currently to do VoIP/Video conferences in Matrix: you instantiate a 'widget', to embed whatever web conferencing service you like into your chat room (e.g. Jitsi). You use the widget API to control it. Widgets aren't actually merged into the spec yet, but are specified at MSC1236 (https://github.com/matrix-org/matrix-doc/issues/1236 https://github.com/matrix-org/matrix-doc/issues/1236), which should get reviewed and merged in real soon now. So, as long as your client implements widgets, you can do conferences. In future, we have plans for native Matrix conferences (MSC2359), but this is vapourware. Finally, if your client doesn't implement 1:1 and doesn't implement widgets (or if your don't have permission to perform them in that room) then obviously hitting the call button won't work. But that's hardly fragmentation; it's just incomplete implementations. Now, if there were multiple competing ways of doing 1:1 calls, or of doing conferences flying around, I'd totally agree we were seeing XMPP-style fragmentation. But we're not, and if we did, we'd see that as a massive bug and jump on it to resolve the fork in the Matrix Spec Core Team, and ensure that matrix.org/docs/spec resolved the collision asap. > Every time Matrix comes up, someone mentions xmpp and immediately the flinging starts about how broken it is and how those developers let it slide into obscurity. Perhaps, but that's not me or the Matrix core team doing that (well, other than back in the very early days when I was still feeling a bit salty about XMPP). We can't control what randoms on the internet say though :|
pseudalopex 6y ago
The approaches are more similar than you think. At least the parts you mention Experimental XEPs let people experiment. The compliance suites provide a single spec controlled by a committee. And the differences don't matter when the user experience is the same. XEP-0027 OpenPGP has been officially obsolete for Matrix's entire life. XEP-0374 OpenPGP never went past experimental. It was officially deferred 3 years ago. OMEMO is the only proposal active. The compliance suites tell you what you need for file transfer. Core requires XEP-0363 for file upload. Advanced requires XEP-0234 and XEP-0261 for direct transfer. You can ignore the experiments. Figuring out what features 2 clients share can be confusing with either protocol. Looking at Matrix's clients matrix[1] is easier than finding a list of XEPs for each client. But you still have to figure out what the server supports and what it doesn't have to. The real difference is how 1 company dominates the Matrix ecosystem. [1] https://matrix.org/clients-matrix/ https://matrix.org/clients-matrix/

petre 6y ago

> Whereas Matrix ensures that there is only ever one way of doing a given operation at any point Python used to claim the same thing when it was a fresh new language, as a reaction to Perl where the opposite was encouraged. Becoming mature and anchored in reality changed some of that.

arendtio 6y ago

> Whereas Matrix ensures that there is only ever one way of doing a given operation at any point I trust you that you mean what you say, but I wonder what an implementation would look like (from a user-perspective), that supports something like OpenPGP -> OTR -> Ratchet. I mean, upgrading security over time is mandatory and when I remember how it evolved for XMPP I clearly do not want an implementation that still has all the draw-backs that OTR had. So I agree that XMPP is lacking some good governance, but I don't see the rivaling XEPs necessarily as the core problem, as many of them had quite some time between the drafts. The bigger issue is that many clients don't have enough developer momentum, so that XEPs that were drafted like 5 years ago are not implemented yet. Regarding implementation bugs: I know XMPP has the compliance suite, but I feel like it doesn't catch bugs and is more of a basic testing for feature compatibility. Does Matrix have some kind of test suite, that simulates physical disconnects, package loss and the likes for testing real client and server implementations? I know that would not be an easy feat, but I wonder what the best way would be to build something like that, to improve the quality of the existing implementations.

pseudalopex 6y ago
OpenPGP could mean XEP-0374 which is newer than OTR. But XEP-0374 went from experimental to deferred 3 years ago. It isn't really competing with OMEMO.
- arendtio 6y ago
  XEP-0374 is very fresh compared to other PGP-related extensions. Just to give an example, 'XEP-0027: Current Jabber OpenPGP Usage' became active in 2002. If it is competing with OMEMO or not might be a separate discussion, but both can be used to utilize E2E encryption of messages.
  - pseudalopex 6y ago
    XEP-0027 became officially obsolete in 2014. I'm giving them the benefit of the doubt they didn't mean anyone has to think about XEP-0027 in 2021. They implied they could break compatibility some day. So I think they're talking about not having 2 current standards. How they see a transition working is still a good question though.

xorcist 6y ago

> Matrix ensures that there is only ever one way of doing a given operation That's not entirely true though. Press voice call in a client and you may end up with an embedded jitsi, some WebRTC thing, or nothing at all. That's exactly how xmpp clients started to diversify. > I'm not trying to spread FUD about XMPP Yeah, the tone could certainly be improved. Every time Matrix comes up, someone mentions xmpp and immediately the flinging starts about how broken it is and how those developers let it slide into obscurity. Which there may be some truth to, but it really leaves a bitter taste. Especially since much of xmpp lives on in Matrix in bridges to other networks. Matrix really got a flying start there, and that's a good thing! Software should be shared and built on. What worries me a little is that the same problems xmpp had, with protocol verbosity and problematic integration with smartphone platforms, is bascially the same in Matrix. The protocol is at least as verbose, and battery management is at least as hard. As a successor to xmpp, one might have expected to learn more from its problems. I'm certainly not complaining however, the more software the better, and the web client is very nice.

Arathorn 6y ago
> That's not entirely true though. Press voice call in a client and you may end up with an embedded jitsi, some WebRTC thing, or nothing at all. That's exactly how xmpp clients started to diversify. This really isn't an example of protocol fragmentation - it's behaving precisely as expected. There is precisely one way to do a 1:1 VoIP/Video call in Matrix: https://matrix.org/docs/spec/client_server/r0.6.1#voice-over-ip https://matrix.org/docs/spec/client_server/r0.6.1#voice-over.... It hasn't actually changed since 2015 when we first added it to the spec, until recently when there's a proposal to improve it called MSC2746: https://github.com/matrix-org/matrix-doc/blob/dbkr/msc2746/proposals/2746-reliable-voip.md https://github.com/matrix-org/matrix-doc/blob/dbkr/msc2746/p.... Now, this is a proposal to patch the spec - it's a Matrix Spec Change. It retains backwards compatibility with the current behaviour, and once the MSC is approved it'll be merged into the spec and the MSC will be discarded. There is precisely one way currently to do VoIP/Video conferences in Matrix: you instantiate a 'widget', to embed whatever web conferencing service you like into your chat room (e.g. Jitsi). You use the widget API to control it. Widgets aren't actually merged into the spec yet, but are specified at MSC1236 (https://github.com/matrix-org/matrix-doc/issues/1236 https://github.com/matrix-org/matrix-doc/issues/1236), which should get reviewed and merged in real soon now. So, as long as your client implements widgets, you can do conferences. In future, we have plans for native Matrix conferences (MSC2359), but this is vapourware. Finally, if your client doesn't implement 1:1 and doesn't implement widgets (or if your don't have permission to perform them in that room) then obviously hitting the call button won't work. But that's hardly fragmentation; it's just incomplete implementations. Now, if there were multiple competing ways of doing 1:1 calls, or of doing conferences flying around, I'd totally agree we were seeing XMPP-style fragmentation. But we're not, and if we did, we'd see that as a massive bug and jump on it to resolve the fork in the Matrix Spec Core Team, and ensure that matrix.org/docs/spec resolved the collision asap. > Every time Matrix comes up, someone mentions xmpp and immediately the flinging starts about how broken it is and how those developers let it slide into obscurity. Perhaps, but that's not me or the Matrix core team doing that (well, other than back in the very early days when I was still feeling a bit salty about XMPP). We can't control what randoms on the internet say though :|

pseudalopex 6y ago

The approaches are more similar than you think. At least the parts you mention Experimental XEPs let people experiment. The compliance suites provide a single spec controlled by a committee. And the differences don't matter when the user experience is the same. XEP-0027 OpenPGP has been officially obsolete for Matrix's entire life. XEP-0374 OpenPGP never went past experimental. It was officially deferred 3 years ago. OMEMO is the only proposal active. The compliance suites tell you what you need for file transfer. Core requires XEP-0363 for file upload. Advanced requires XEP-0234 and XEP-0261 for direct transfer. You can ignore the experiments. Figuring out what features 2 clients share can be confusing with either protocol. Looking at Matrix's clients matrix[1] is easier than finding a list of XEPs for each client. But you still have to figure out what the server supports and what it doesn't have to. The real difference is how 1 company dominates the Matrix ecosystem. [1] https://matrix.org/clients-matrix/ https://matrix.org/clients-matrix/