3 ms·

> fragmentation is the biggest thing holding back desktop Linux I'm not convinced. The main thing holding back desktop Linux is buggy upstream software not wo

by nh2 2y ago

> fragmentation is the biggest thing holding back desktop Linux I'm not convinced. The main thing holding back desktop Linux is buggy upstream software not working, which usuall affect all distros. LibreOffice Writer didn't have a rastered ruler for 10 years, no quality GPU driver for some AMD card, some KDE app segfaults, some Python app has a value unexpectedly being None, a kernel minor version update just completely broke the version of VirtualBox on Ubuntu 22.04 LTS -- those were and often still are daily problems that no amount of distro hopping or fragmentation consolidation will fix. The "large" distributions justify their existence by differing in key aspects that cannot be consolidated: * Corporate vs community, e.g. RedHat/SuSE/Ubuntu vs. Debian * Fundamental concepts, e.g. NixOS/Guix vs. everything else * Up-to-date-ness, e.g. Arch vs. Debian There are some aspects of fragmentation that _could_ be consolidated by better software and more configurability, e.g. * Source vs. binary, e.g. NixOS provides both * Choice of libc, e.g. NixOS ships glibc by default but you can easily get much software with musl, and it's technically possible (but needs more community power) to provide a whole system running musl with a single line of config for the end user * GUI vs CLI configuration For the small distros, they are niche enough that them merging with another distro doesn't have a huge consolidation effect overall. For example, I do not see how Void Linux and Alpine Linux merging would significantly improve the life of the mass of Ubuntu users, or signficiantly advance "Linux on the desktop". It's true that there's some amount of duplicate work occurring, but much of that is often easily re-used across distros. For example, if in NixOS something doesn't work with the newest version of musl, one would usually check if Alpine already solved that and if yes, swiftly copy the approach. The main work is for the _first_ one to figure it out, and that work is necessary even if there was only 1 distro. Reducing distribution fragmentation isn't what makes "Linux on the desktop" happen. Fixing broken software, making good desktop environments, creating Wine or Proton or Pipewire or cross-platform software (LibreOffice, all browsers, Electron), does.

al_borland 2y ago
I don't think we disagree at a fundamental level. My thought on reducing fragmentation is to reallocate those resources to fix the broken software and to make everything better and smooth. That last 10% to make the UX really good isn't something most people like doing, but it's something that macOS and Windows both spend a lot more time on that Linux. Due to the fragmentation, I don't think most users make it far enough to run into the application or UX issues. It seems like reducing fragmentation would get more people into the door, and once they are in the door, having those developers work on making the core experience better, is what keeps people around. Once the platform has more users, that will bring in even more developers... the ones who are there as a job, not just enthusiasts, which should lead to an upward spiral for quality. In terms of the large distributions not being able to be consolidated... I don't think the average user (read: typical Windows and macOS users) care about any of that, other than not having to go to the CLI to change a config. A user should not need to know what glibc is.
- diffeomorphism 2y ago
  Entirely different resources though, which is why this criticism makes little sense. It is like asking management to help coding the backend -- no, you need different people for that.

by nh2 2y ago

al_borland 2y ago

I don't think we disagree at a fundamental level. My thought on reducing fragmentation is to reallocate those resources to fix the broken software and to make everything better and smooth. That last 10% to make the UX really good isn't something most people like doing, but it's something that macOS and Windows both spend a lot more time on that Linux. Due to the fragmentation, I don't think most users make it far enough to run into the application or UX issues. It seems like reducing fragmentation would get more people into the door, and once they are in the door, having those developers work on making the core experience better, is what keeps people around. Once the platform has more users, that will bring in even more developers... the ones who are there as a job, not just enthusiasts, which should lead to an upward spiral for quality. In terms of the large distributions not being able to be consolidated... I don't think the average user (read: typical Windows and macOS users) care about any of that, other than not having to go to the CLI to change a config. A user should not need to know what glibc is.

diffeomorphism 2y ago
Entirely different resources though, which is why this criticism makes little sense. It is like asking management to help coding the backend -- no, you need different people for that.