Back when the GNU project was starting, among the first things they rewrote as Free Software were:
- text editor / IDE (Emacs)
- assembler, linker, and compiler
IOW, they made tools that they needed to further develop Free Software without relying on proprietary tools.
They wanted their project to be self-hosting.
Nowadays, we have more free software than ever, but we develop it using github and Discord...
@Wolf480pl this is issue with people. People tend to stich with a single solution which everybody uses. There are lots of quailty source "forges" (i.e. Gitlab, Gitea, SourceHut), but people stick with GitHub for a reason that "everybody uses GitHub".
At the moment, when everybody will use open and free forge. Especially when the #ForgeFed will be adopted.
But it is harder instant messagging systems. Every one of them has one issue that makes it harder to use.
IRC - lack of history *on server*. I don't see reason to chat if the history is not saved even for several minutes.
XMPP - it's high entrance point. Too much clients and servers supporting different featuresets.
Matrix - it's pretty young and suffers from it.
RocketChat - AFAIK it was buggy as hell, when last time I was using.
Gitter is very tight coupled with non-free software
Either way, it's clear that these days we need something better than that.
My problem is that instead of being like "IRC is no longer sufficient for our project, let's make something better", many FOSS projects are like "IRC is no longer sufficient for our project, let's use some proprietary chat app instead".
Which is the opposite of what GNU was doing back then.
Where I live, there's an expectation that "everyone is on facebook", and when you're that one guy who is not on facebook, you're basically a second-class citizen.
Please don't teach your friends to expect that everyone uses the same communication medium.
Making a system that is decentralized, secure, and lightweight (in terms of network usage) is an extremely difficult thing to accomplish; most projects or protocols only place value on two of the three as a result.
@alexcleac @jack @Wolf480pl It’s a general principle. Reliability costs performance, and security requires trust analysis which incurs overhead. I can process half a billion transactions a second in a system written in assembler that doesn’t care about networking or security. That doesn’t make it good.
But how much you want to bet that it’d still sell well? “Half a billion txns/sec on a four year old system! Buy now! Be faster than everyone else!”
This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!