I recently started using #ElementX as my matrix client, which required me to set up a separate "syncv3" service alongside synapse on my homeserver. Works great.
It's my understanding that Element X is a re-write that will eventually replace mainline "Element" in the app stores. What happens then to Element users who aren't using a sliding-sync-enabled homserver? Does it just break? Is sliding-sync going to be folded into Synapse/Dendrite? If so will syncv3 be deprecated?
I don't know what the organizations' issues were, but when I use IRC, I'm typically idle in several rooms and only rarely post anything. That doesn't mean I'm not reading the contents regularly. But it does mean that the bridge would terminate your session and you'd have to go through the whole process of connecting to the desired rooms from scratch.
I found this burdensome and incompatible with the way IRC is used, so I decided some time back that I wasn't going to use Matrix as an IRC bouncer any more. (Which also means that most of my use of Matrix has ended. I'm in several rooms where there were less than 10 posts total so far in 2023, and now I find myself not checking regularly ... and naturally notifications are turned off.)
@clacke Yes. That isn't only true of #Discord, by the way. I used to see projects nearly devoid of documentation, but they'd push you to #IRC. And same for #Matrix or #XMPP MUCs.
I use IRC, but if I need info (a config setting, the setup process, etc) some questions recur and should be documented so they don't clog up the channel. Others occur so rarely that only one or two people know a fix. Those especially need to go into the documentation. Not just because of "truck factor", but because even those people can forget if there is a long enough time between the question recurring.
Afterwards, I did a test connection to the #Federati #Murmur / Mumble server. It disconnected once over a ten minute test.
So if I do another #Office_Hours voicechat soon, understand that my connection may be interrupted.
Anyway, I got some great ideas. A has been trying to document #ActivityPub itself (S2S and C2S designs), various implementations (including the Mastodon API that is often used instead of AP_C2S), as well as other #federated communication protocols such as #Matrix and #XMPP. They suggested that "a better Matrix" might be able to be built atop AP.
I used to watch their game streams years ago, but the voice sounded different over Mumble. I'm told that getting a better microphone made the difference.
They note that they provide everything under permissive licenses and then wonder why giant corporations adopt their technology without giving anything back.
Well, that is something that MongoDB and some other no-SQL databases have had to deal with. Their userbase insists on permissive licenses, then the largest and most profitable users refuse to support the organizations that pay the developers. In many cases, the result is some custom non-free license (generally masquerading as “open source” but neither free software nor open source software) meant to extract money out of users who make money using the product.
Personally, I’d rather see #GPLv3 / #AGPLv3 without a “buy a commercial license to avoid responsibilities” plan going into widespread use, but that’s sure to greatly reduce revenues. It probably means paid support is the main funding model.
But anyway, I think that Matrix has done really well for something that is inherently inferior to #XMPP, but has a pretty face and good marketing team.
I do use #Matrix a little, but I have been saying all along that I don't like the way they bolted on end-to-end encryption (E2EE) after the protocol was already in use. There are constantly issues (see the \#element-ios:matrix.org and \#element-android:matrix.org rooms for regular traffic about issues related to encryption and the "verify my device and session" process that should never appear. And there is no concept of a contact ... there is some simulation in the way one-to-one chats are labelled, but there are some weird things that can happen if one expects that a list of those chats are a contact list.
Some friends and I have a private Matrix room that we've been active in daily for several years. I've had the thought that it would be fun(ny) to train a bot using our chat history and have it simulate us talking to each other in a dedicated room. However, I have almost no practical knowledge of how to implement machine learning or language processing. Can someone point me to some resources a noob could use to get started? I'm very comfortable in python.