Lennart Poettering Talks Up His New Linux Vision That Involves Btrfs

[nilssab]

inb4 btrfs becomes the required linux filesystem.

want dbus? you need btrfs!

Disclaimer: I use and love btrfs and Lennart is my personal hero after Linus and Greg.

I can't wait for that to also follow into floods of support threads from casual users on 64gb ssds that cant seem free up space on their drive

[Sonadow]

[quote=Lennart and his fellow systemd core developers have been devising a new way to tackle these challenges to come up with an efficient way to allow vendors to package their software for end-users, allow end-users/administrators to install packages regardless of the distribution at play, a unified solution to cover everything, [/quote]

Good luck trying to unify all the differences between every single distribution on the planet.

Especially those like

- *-dev vs *-devel
- /lib32 + /lib (pre-multiarch Debian-based, Gentoo and Arch) vs /lib/<arch> (Debian-based multiarch) vs /lib + /lib64 (rpm-based)
- different distributions installing certain packages in different locations (at one point OpenSUSE installs KDE into /opt, not sure if they do this now)
- different naming conventions, especially for libraries (e.g.: Mageia / Mandriva uses libfoo and lib64foo, Fedora uses libfoo and libfoo.i686, OpenSUSE uses libfoo and libfoo-32bit)

[kshade]

This sounds interesting, worthwhile even. I don't trust that particular group of developers not to make it an overly complicated, incompatible mess though. We'll see.

[Awesomeness]

Good luck trying to unify all the differences between every single distribution on the planet.

Then they stop the madness and become Fedora and CentOS remixes.

[asdfblah]

experimental init system on top of experimental file system?
Take my money!! /s

[Awesomeness]

This sounds interesting, worthwhile even. I don't trust that particular group of developers not to make it an overly complicated, incompatible mess though. We'll see.

btrfs was conceived by Oracle and is now maintained by Facebook, ie. the two most popular companies among Linux users. Nothing could go wrong.

[Awesomeness]

experimental init system on top of experimental file system?
Take my money!! /s

systemd is not experimental. RHEL7/CentOS-7 uses it and it works just fine.

[Brane215]

I like his views in general and this idea in particular.

Endless clinging to least common denominator is so limiting and pointless. Why develop usefull new features if they are going to be used just for the show ?

Although I don't really trust btrfs yet, if one is serious about making modern solutions, one would be wise utilize to the full features available.

And btrfs has a few that would make life easier for user and maintainer. Yes, Linux has 30 or so filesystems, but all that "choice" takes more through fragmentation than it brings in freedoms.
So, it could be the right time to place btrfs to the fore and push extX back.

Also, if btrfs is ever to be seen as serious choice, it has to take training wheels of. Creating interest only help to build a pressure needed for polishing remaining rough edges, fixing last important holes and creating good fsck tools.

After that, mainstream Linux will have its FS of choice and some standardised way of providing modular support for virtual machines, work in containers or just different component combinations of host OS.

[dsx724]

Meh

Phoronix: Lennart Poettering Talks Up His New Linux Vision That Involves Btrfs

Lennart Poettering of systemd and PulseAudio fame has published a lengthy blog post that shares his vision for how he wishes to change how Linux software systems are put together to address a wide variety of issues. The Btrfs file-system and systemd play big roles with his new vision...

http://www.phoronix.com/vr.php?view=MTc3NzU

If you scroll to the bottom where he summarizes the proposal, I came to the inevitable conclusion that we won't need any of the points he is making because there is either already a way to do it, there is no necessity for doing it, and/or it's impractical to do it. Lennart Poettering needs work in the real world for a while and get out of his bubble. Given infinite time and money, you can do it but the gain is not justified for the time expenditure.

Systemd is a mistake and goes against a few decades of wisdom of open source development. Adding even more dependencies to Lennart Poettering Linux(TM) is his prerogative. Forcing Lennart Poettering's stack into every Linux distro is a mistake.

[smitty3268]

It's an interesting concept, but...

It's going to be tough to get all the different distros to agree on a common way of doing things like this. Really tough. Not sure i see it as feasible or not, but i guess they're free to try. And I suppose they could just do it on Red Hat/Fedora themselves if they really wanted to - nothing really requires other OS's to join in.

This also allows us to implement something that we like to call Operating-System-As-A-Virus.

That's certainly not going to get taken out of context, is it.