This is some exciting technology, I wonder why they are not highlighting why it doesn’t work in Private Mode (I guess because WebRTC is disabled there). Also seeing an sqlite database with a go-sqlite-js through WASM in your browser definitively gives some living in the future vibe.
Firefox currently blocks access to IndexedDB in private browsing mode, and the P2P client uses this for storage as it runs. Looks like there’s a recent effort to fix this on the Firefox side.
It’s Dendrite based, so I have doubts about the sustainability due to New Vector’s constant changes on Synapse breaking compat with everyone else; another sad manifestation of the perverse incentives preventing Matrix from being a stable and performant alternative.
Wow, this is some quality FUD. We haven’t broken backwards compatibility ever on the Client Server API - you can spin up a Matrix client written pre-launch in 2014 and it should work fine with a 2020 Matrix server. On the Federation (Server-Server) API, we upgrade the room synchronisation protocol on a semi-regular basis using room versions (https://matrix.org/docs/spec/#id5)… and we’ve still kept backwards compatibility with the earlier room versions.
For wider context: New Vector (NV) is the company set up by the original Matrix team which sponsors much of the development of the core Matrix project. The accusation here is that New Vector is somehow incentivised to sabotage development of Synapse (the reference Matrix server) to prioritise its own commercial interests at the expense of the wider Matrix network. This is categorically untrue. All work that NV people do on Synapse is donated unilaterally to the Matrix.org Foundation, which is governed incredibly strictly (including by the UK non-profit regulator) to ensure the protocol and reference implementations advance for the benefit of everyone - without ever prioritising any commercial interests, particularly NV (now or in the future). https://matrix.org/foundation spells this out.
So: a) Synapse doesn’t break compat; b) NV doesn’t prioritise itself when doing Matrix.org development, and if it did, the Foundation would course-correct; c) Dendrite work is mostly funded by NV anyway :/
This is the impression I got from observing the Matrix project, Dendrite’s progress specifically over Synapse, and third-party implementations. The biggest pain point is how checking off features seems to be prioritized over actual polish and optimizations to the servers themselves.
I think you are extrapolating incorrectly. Almost all work in Synapse over the last year or so has been around polish & optimisations - and Dendrite dev was on hold in order to focus on Synapse stability & perf. In the last few months we’ve been able to afford to spin up Dendrite dev again, and meanwhile Synapse is (for once) stable. We’re almost at the point of going back to feature dev - but other than E2EE-by-default, there’s been very little feature dev since 2018; it’s all been about stabilising for Matrix 1.0 (in June 2019) and then subsequent polish.
I think you might be basing your conclusions on where things were at in 2018 or so (when we had indeed been rushing features out the door in order to try to secure long term funding for the project, and to try to keep up with the centralised competition).
NV is the startup which the original Matrix team formed in 2017 in order to fund themselves to continue working on Matrix fulltime. Of the core Matrix.org codebase, probably ~90% of it is written by NV employees, all of which is donated to the Matrix.org Foundation, which was set up as an independent custodian of the protocol itself (protecting it from NV and any other commercial players). The wider ecosystem building on Matrix spans hundreds (thousands?) of various projects, projects and contributors.
So it’s not coincidence that NV drives a lot of Matrix development, given all but one of the team who created Matrix work there :)
I can’t follow the reason for your criticism without further explanation. What breaking changes did Synapse introduce in the past? Who is “everyone else”? What kind of “perverse incentives” do you mean? Why should this be a problem for Dendrite, which is also developed by New Vector? Any sources?
New Vector is basically paid to develop features over stabilizing and polishing the low-level details. This made federation somewhat of a moving target, and it means things like performance are moved to the wayside.
It constantly amazes me that the only thing that a lot of people like to complain more than proprietary software is free and open source software that is done by people other than themselves.
This is some exciting technology, I wonder why they are not highlighting why it doesn’t work in Private Mode (I guess because WebRTC is disabled there). Also seeing an sqlite database with a go-sqlite-js through WASM in your browser definitively gives some living in the future vibe.
Firefox currently blocks access to IndexedDB in private browsing mode, and the P2P client uses this for storage as it runs. Looks like there’s a recent effort to fix this on the Firefox side.
Thank you for this information, very much appreciated.
Or past if we consider websql(ite).
It’s Dendrite based, so I have doubts about the sustainability due to New Vector’s constant changes on Synapse breaking compat with everyone else; another sad manifestation of the perverse incentives preventing Matrix from being a stable and performant alternative.
Wow, this is some quality FUD. We haven’t broken backwards compatibility ever on the Client Server API - you can spin up a Matrix client written pre-launch in 2014 and it should work fine with a 2020 Matrix server. On the Federation (Server-Server) API, we upgrade the room synchronisation protocol on a semi-regular basis using room versions (https://matrix.org/docs/spec/#id5)… and we’ve still kept backwards compatibility with the earlier room versions.
For wider context: New Vector (NV) is the company set up by the original Matrix team which sponsors much of the development of the core Matrix project. The accusation here is that New Vector is somehow incentivised to sabotage development of Synapse (the reference Matrix server) to prioritise its own commercial interests at the expense of the wider Matrix network. This is categorically untrue. All work that NV people do on Synapse is donated unilaterally to the Matrix.org Foundation, which is governed incredibly strictly (including by the UK non-profit regulator) to ensure the protocol and reference implementations advance for the benefit of everyone - without ever prioritising any commercial interests, particularly NV (now or in the future). https://matrix.org/foundation spells this out.
So: a) Synapse doesn’t break compat; b) NV doesn’t prioritise itself when doing Matrix.org development, and if it did, the Foundation would course-correct; c) Dendrite work is mostly funded by NV anyway :/
This is the impression I got from observing the Matrix project, Dendrite’s progress specifically over Synapse, and third-party implementations. The biggest pain point is how checking off features seems to be prioritized over actual polish and optimizations to the servers themselves.
I think you are extrapolating incorrectly. Almost all work in Synapse over the last year or so has been around polish & optimisations - and Dendrite dev was on hold in order to focus on Synapse stability & perf. In the last few months we’ve been able to afford to spin up Dendrite dev again, and meanwhile Synapse is (for once) stable. We’re almost at the point of going back to feature dev - but other than E2EE-by-default, there’s been very little feature dev since 2018; it’s all been about stabilising for Matrix 1.0 (in June 2019) and then subsequent polish.
I think you might be basing your conclusions on where things were at in 2018 or so (when we had indeed been rushing features out the door in order to try to secure long term funding for the project, and to try to keep up with the centralised competition).
What % of Matrix development is funded/driven by New Vector? It’s cool that they are pushing for this kind of software.
NV is the startup which the original Matrix team formed in 2017 in order to fund themselves to continue working on Matrix fulltime. Of the core Matrix.org codebase, probably ~90% of it is written by NV employees, all of which is donated to the Matrix.org Foundation, which was set up as an independent custodian of the protocol itself (protecting it from NV and any other commercial players). The wider ecosystem building on Matrix spans hundreds (thousands?) of various projects, projects and contributors.
So it’s not coincidence that NV drives a lot of Matrix development, given all but one of the team who created Matrix work there :)
I can’t follow the reason for your criticism without further explanation. What breaking changes did Synapse introduce in the past? Who is “everyone else”? What kind of “perverse incentives” do you mean? Why should this be a problem for Dendrite, which is also developed by New Vector? Any sources?
New Vector is basically paid to develop features over stabilizing and polishing the low-level details. This made federation somewhat of a moving target, and it means things like performance are moved to the wayside.
It constantly amazes me that the only thing that a lot of people like to complain more than proprietary software is free and open source software that is done by people other than themselves.