@david_chisnall I feel that there is a tension between integration and coupling. On one hand, tightly integrated systems tend to work really well together. Yeah, I know, big surprise there. But on the other hand, loosely coupled systems tend to be really easy to take apart and modify.

I am not sure whether we can get both.

@david_chisnall

"These components are large monolithic entities.
[..]
And each of these projects is too complex for a single person to create a replacement for unless they treat it as a full-time job."

By Design: Those seeking to reduce
the explosion of diversity protection against least cost unwarranted penetration.

In that sense, it is a winner. The Organisations hosting the project get to call the shots for security "best practice" too. win-win.

@david_chisnall
Here, I fixed it for you: https://distrowatch.com/search-mobile.php?ostype=Linux&category=Desktop&origin=All&basedon=All&notbasedon=None&desktop=All&architecture=All&package=All&rolling=All&isosize=All&netinstall=All&language=All&defaultinit=Not+systemd&status=Active#simpleresults

Now seriously, I get where you're headed and agree that some software is aggregating too much, but those maybe are not the best examples for 2, as alternatives exist.
And this definitely doesn't say much about FOSS, it says about Linux demographics changes. FOSS is independent of the UNIX philosophy, although they are perfect together.

@david_chisnall That just sounds like "Not Invented Here". The thing is easy to fix as long as it is somebody else that does the fixing for us.

It is also minimizing the hard work of some of the most brilliant minds of our time.

@david_chisnall I'm writing a shell I talk about way too much that I'm pretty excited about for very much this reason

@david_chisnall I agree with you on the following points:
- there are open source projects which do not prioritize good system and software architecture enough
- there are open source projects that do not prioritize (and sometimes do not care one single bit) about customer value
I do not agree on the general notion or what I understand. #systemd and #wayland are not something chosen regularly by end-users but by engineers from distributors.

@david_chisnall Exactly. The complexity makes 'just fork it' unrealistic for most. We need simpler core components.

@david_chisnall

> To me, that really highlights the failure of the Free Software movement.

I missed anything that would lead me to agree with the idea FOSS has "failed", as read your claims on the FOSS mastodon, and write this reply on a 100% FOSS system on wayland + systemd. It all appears to be doing great?

Even the complaint this stuff others spend their time on the planet on for you to use for free is "hard" may start to buckle since you can ask your AI buddy to do the change for you.

@david_chisnall This describes perfectly how I feel about those projects, I was just never able to put it that succinctly. Thank you!

@david_chisnall To be fair, this is a direct consequence of how a large part of software in general is developed over time: by aggregation, meaning that a project that starts small and maintainable at the beginning aggregates features that increase complexity, but this does not lead to a redesign to reduce this complexity, because that would make the current users and developers unhappy and that splitting functionality into small loosely coupled parts is hard even when done from the beginning.

@david_chisnall
First OS I ever used I wrote in two days using machine language. I next wrote a simple ed like editor, and then an assembler. That was back in the mid 1970s.

I've watched systems get more and more complex and layered over the decades. I expect Evolving Intelligent systems will take over coding and user interface. It will take a couple decades to get them right. At that point, it won't be coding, but the knowledge of how to specify that will matter.

@david_chisnall IMO, a very strong factor are companies coopting development and turning projects into monolithic products... Google, Valve, Canonical or Red Hat, among many others, are users of Linux and its ecosystem, but they have the resources to pay key developers, and their influence on the design is disproportionate compared to other users. Usually their contribution is crucial, but it comes with strings attached. In an ideal world, these companies would just send money, not MRs

@david_chisnall I could not agree more. Linux won, but it is pyrrhic victory. As someone who came to Linux as a teen, the UNIX philosophy was truly empowering. You could take apart every part of the system and adapt it or replace it. And with open-source for each component this was truly enabling. Over the last decades, Linux is evolving away from this and nowadays I often need much more time to adapt some aspect (or just revert something upstream changed to my disliking) than I can afford.

@david_chisnall that would mean to flip open source development into protocol standardization ??

like in "protocols, no platforms" ???

@david_chisnall I don't do much coding these days so I probably shouldn't comment, but I'd willing to take a crack at reimplementing a systemd component if I had to - the protocols etc seem relatively simple. Absolutely not with anything graphics-y, though :(