These are my answers to @retiredt00's questions. Anyone who shares my view is welcome to contribute to https://github.com/vanilla-dpup/woof-CE. I will accept contributions that follow this spirit, and won't accept messy, complicated or out of scope things that become a maintenance burden.
If you have a different vision, go on and make it happen. Puppy is a do-ocracy. My work is not supposed to replace upstream woof-CE in any way, only create a parallel branch of development that's based on my opinions.
---
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
Bleeding edge kernels of stable longterm security-supported only?
Both. A stable version with several years of security updates and low-risk fixes for high-severity bugs, plus a bleeding edge development version.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
What is the evidence that puppy users get a newly build machine and peripherals every 6 months?
Who (what percentage) uses puppy in a production system?
I see more and more screenshots of computers with 16 GB of RAM. Therefore I think Puppy should be built to do "progressive enhancement": for example, automatically free RAM used by SFSs copied to RAM in a low RAM situation, but try to maximize performance at the cost of RAM consumption when possible (like most OSs do).
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
The more flexible Aufs or the in-tree overlayfs?
overlay.
The Vanilla Dpup 9.x line has been consistently shipping bug fixes and security fixes on top of the 5.10.x kernel series for over 2 years (!!!) and this is not possible with aufs because it breaks often and drops support for still maintained "longterm" kernel branches way before they reach their EOL date.
Using aufs in a distro for old computers or a stable distro is a very bad idea because it forces frequent kernel upgrades, to chase the limited range of kernel versions supported by aufs. Newer and newer kernels support less and less old hardware, and add new issues while "longterm" kernels age like wine and gain only fixes.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
Modern Wayland or more stable (for the moment) Xorg ?
Wayland, because X.Org is dying. Latest X.Org fails to start on my oldest laptops and Wayland doesn't work either, so I don't see a reason not to drop support for X.Org and just migrate to Wayland. Most users of major distros already use Wayland, and I've seen cases where applications that work only thanks to Xwayland (= not ported to Wayland to run natively) no longer work under X.Org because nobody runs then under X.Org anymore.
Support for old hardware in the modern Linux graphics stack (maintenance of the Mesa Amber branch, etc') is not a Puppy-specific concern and those who want to support old hardware should do some upstream work first and their changes will eventually propagate to Puppy. Somebody needs to do this work, and the expertise required is not the same as the expertise required to build a distro.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
All possible desktop and file mangers or limited resources/compatibility cautious selection?
By default, a specific Wayland compositor, one file manager, etc', that are preconfigured and patched to make them familiar to existing Puppy users. Users should be able to switch to something else, but it takes a lot of work to achieve the same degree of cohesiveness and polish.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
Importantly, what puppy-specific applications/utilities/managers will be included, if any, to distinguish puppy from other debian/slackware derivatives?
1. Many yad-based configuration tools and wrappers for tools like wf-recorder which force the user to use the terminal
2. A lightweight and fast init system that reduces the number of running processes
3. Various tweaks for low resource consumption and increased longevity of flash storage
4. Various security and privacy tweaks - ad blocking and firewall that are enabled by default, various browser privacy tweaks, and support for encrypted or partially encrypted save folders
5. ... everything in https://github.com/vanilla-dpup/woof-CE ... e#features and more, as more features are added
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
All the distributions based on the 4-5 main ones add a lot of customization in look, feel and utilities to make their mark.
What puppylinux will have on that front?
Very little customization and artwork because many applications don't want to be themed, and changing the appearance of GTK+ 4 applications requires patching of libadwaita. The distros that do this (like Ubuntu) make very subtle changes anyway, so IMO the big effort is not justified in a small community distro with few developers.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
More important who/how is going to maintain all these puppy-specific apps?
The same people who maintain the build system. Everything is in one place, it's a single git project.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
Build from source, with modified distribution packages or with original distribution packages?
Both. Puppy-specific stuff is built from source by the build system and the source code is part of the build system, the rest comes from binary packages.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
Will building (and testing that most distributions do now) be automated or hand crafted and alpha-user tested
100% automated and reproducible.
retiredt00 wrote: ↑Tue Aug 06, 2024 5:57 am
However, the answers to all these and more must address the basic question which is "use case".
The original Puppy use case, a lightweight and fast distro people can use for daily tasks. While some in-house tools are crude, simple and amateurish, they get the job done.