Hacker News Reader

Comment by knocte | Hacker News Reader

Keep making the biggest problem of Linux bigger, please... Yeah, this problem is "fragmentation", and it is not solved with "simplicity".

View on HN

This kind of distro is exactly the point of FOSS & Linux imo. Diversity of ideas breeds innovation and competition, we need that. Otherwise linux will at some point go the way of browsers, with everything effectively owned by google/microsoft/whatever and a big fat artificial lock-in.

In addition, this kind of project is emblematic of the hobbyist nature (and origin) of linux. When that's no longer possible, the fun and interest, for me at least, is dead. And when I get gently pushed into some kind of software standardization I can't opt out of, I'm going return to windows.

You are right ! diversity is the good move, that is how life decided to grow, by diversifications. Ok, it consumes loads of time, but the result is without point of thinking. M$ did the opposite in '80, and is not quite winning nowdays... Just a question of time and experimentations ;)

No, that's not the point of FOSS. Having a right doesn't mean that you have to exercise it for the sake of it. When I use/advocate Linux and other FOSS apps it's because ultimately I'll be able to fix any issue that they have, or that nobody will be able to take it the software away from me. But I'm not creating patches for every piece of software I use. We developers should grow past that point, the reason why so many ISVs don't support Linux is because of this madness which means there's no really standards in this ecosystem. (Yeah, now cry me a river about how standards are the equivalent of "authority" please...)

Really, there are de-facto standards. If you are a big corporate enterprise, like a bank, the standard is going to be RHEL. If you're conservative organization but don't want to pay for licenses or support, you're probably going to go with a RHEL clone like CentOS, or worse, Oracle Linux. If you're a trendy Web 2.1 startup you'll probably go with Ubuntu or containers of Alpine Linux etc.

EDIT: source - been doing Linux for corporates, big and small since 1999.

They fragment on trivial parts which is pointless for the evolution.

I think comparing software development to a barely understood natural phenomenon like evolution is a slippery slope, but assuming we take that position, evolution has produced many biological dead ends through mutation. So why should software be any different, it is produced through a biological process of monkeys hammering on keyboards after all.

Your comment is the best illustration of what fragmentation means.

Then choose Windows then. Or Solaris. Or any of the other major operating system players. It really doesn't matter, unless you consider value for money or being able to get things done as important.

Having multiple choices for different purposes is completely normal, even for Windows - or would you install Windows 10 Home on an enterprise server or an IoT device?

You call it "fragmentation", others might call it "choice".

The main problem with Linux adoption isn't choice - every beginner can easily find out a reasonable distro for starting out - but motivation to do the switch in the first place. Almost nobody is used to Linux and doesn't know a reason to try it. Even limiting the choice to one single distro isn't going to change that.

This "problem" you talk about is not a problem in my view but rather a essential freedom in free software.

People's computers are like their cars we like to customize and mod them because it's cool and remember, you own it so it gives a sense of pride.

My problem with the proprietary software is that they compete by creating incompatible standards for the same thing to fight each other while I'm suffering implementing them or using them and having to jumping from one to the other.

There was a glimmer of hope at the start when the focus was on just getting the full free software stack working. But then it went back to making near-exact copies of the most trivial parts of it.

Why nobody ever copies the hard parts?

If you want to get rid of "fragmentation" to make linux "win", you get rid of a big part of what makes linux "winning" a desireable goal in the first place.

How is fragmentation a "big problem"? There are like what, four distros?

- Debian and *buntu - Red Hat - Fedora - Suse

All of them use systemd. All use X11 or wayland. All package gnome. All DEs, WMs etc. are available on all these distros.

Smaller distros can reuse whatever is available for that distros, see for example chrome or skype.

What do you mean reuse whatever is available? Non derivative distros are going to be using their own package manager and not just using some prebuilt package for another system.

But regardless, I agree, fragmentation isn't really a problem. It would be very rare indeed for some software to only work on some distro but not another. And if this was the case, you could probably change it with enough work (swap out libc or whatever).

I mean for example this:

https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=googl...

Google only packages chrome as .deb and .rpm, but you can still get it on other distros anyways.

This isn't going to cause any meaningful fragmentation because, simply looking at its goals and features, very few people will find it usable at all, and those who do will be an extremely small set of highly technical users.

I think not to worry about it, it will probably die when the maintainer gets bored of it after a few years. You don't have to use it. And in the meantime, somebody gets to learn to more by doing their own LFS.

I think that distributions patching software is especially problematic -- whether you consider it a symptom or a cause of fragmentation.

You're missing the point of free software. You own it and can change it how you like. It's not problematic this is how it was made to be.

Here's two examples of how distros patching software has been helpful for me.

1) SquirrelMail (abandoned) patched to work with PHP 7.3 thanks to FreeBSD contribs

2) Abiword patch builds in the AUR that fix a broken default install.

Is the point of free software to create as much variations of a program as possible?

I'd rather see that useful changes are merged upstream, or projects are forked. If this had happened, you wouldn't even need to patch, and this would save an enormous amount of manpower that is wasted on trivialities.

Distributions exist to provide software that all fits and works together (e.g. providing the correct versions of dependencies). Sometimes patching is necessary to make things work.

Why would it be a problem? I think it's a strength.

Reply on news.ycombinator.com