ex-Debian contributor discovers squashfs-based package management

[GNUser]

Michael Stapelberg does research on package managers. He came to the conclusion that squashfs-based package management works best. He also recommends keeping packages as small as possible, with minimal metadata and no included man pages. He also discovered that keeping packages read-only rather than installing them "scatter mode" all over the file system prevents bit rot. Does any of this sound familiar?

His Eureka moment: "Complexity is not inherent: I can build a Linux distribution for general-enough purposes which is much less complex than existing ones."

https://michael.stapelberg.ch/posts/2019-08-17-introducing-distri/

After a few more years of research and polishing his package manager, I have a very good idea what his end product will look like (cough, tce, cough, cough).

[Rich]

Hi GNUser

... Does any of this sound familiar? ...

I feel like I've just experienced a case of deja vu. 

[GNUser]

The June 2021 edition of Linux Pro Magazine has an article on Stapelberg's distri project, which is how I found out about it. 

Alpine was mentioned as having the fastest package manager of those he tested, followed by Arch, followed by everybody else, followed by Debian and Fedora.

I guess Stapelberg never tried TCL. When he does, it will rock his world.

[GNUser]

Here are results of adding a relatively large application (ffmpeg) to two comparably bare-bones installations, both fetching from my local mirrors, using same antediluvian hardware:

Code: [Select]

TCL11 x86_64$ time tce-load -wi ffmpeg4
---snip---
saving to 'ffmpeg4.tcz'
ffmpeg4.tcz          100% |***************************************************************|  196k  0:00:00 ETA
'ffmpeg4.tcz' saved
ffmpeg4.tcz: OK
real 0m 26.71s
user 0m 3.22s
sys 0m 1.89s

Code: [Select]

Alpine x86_64$ time sudo apk add ffmpeg
---snip---
(20/21) Installing ffmpeg-libs (4.3.1-r3)
(21/21) Installing ffmpeg (4.3.1-r3)
Executing busybox-1.32.1-r6.trigger
OK: 445 MiB in 156 packages
real 0m 34.64s
user 0m 1.48s
sys 0m 0.70s