MZan

An inspirational reply of koverstreet https://lwn.net/Articles/1028572/ He is focused on solving all bug reports, and in iterating bcachefs in the faster way possible. It is a sort of implicit contract with its users base: “you are using bcachefs in production/real-world despite it is experimental, and I will support you restoring files and improving code”.

So, I can understand why he doesn’t like waiting too much before releasing improvements. The Linux kenernel release cycle can transform weeks in months if some bug-fixes requires new features or refactorings, like in case of a new file system. I’m sorry that him and Linux maintainers didn’t find a good approach.

BTW, I used bcachefs for 1-2 years with 3 HDD and 1 SSD in cache. It supported this usage scenario better than ZFS for a desktop/workstation like mine setting.

Bcachefs was not kicked out because its design and implementation is of bad quality. A bad quality software is something that is: full of bugs; every 4 fixed bugs you introduce a new one; the design is so wrong that it is difficult to add new features; the code is difficult to understand.

If you consider various metrics like irrecoverable loss of data, lines of code, features, Bcachefs is a very good project. It is under development, so the source code is not “stable” and there are frequent updates, but the behavior of the file system is fully under control and it is only improving,

Bcachefs seems a really well designed filesystem, because it takes in consideration the lessons learned from other filesystems and its code base is compact and elegant.

It is also very robust because corrupted metadata can be reconstructed scanning the leaf nodes with data. So there are very few weak points in case of disasters. Infact every time there were a report of corrupted data, the bcachefs team was able to fix the code and retrieve the data of the affected users.

I used it for a while, and I were impressed. For example it automatically upgrade its data structures.

It is true that there were a commit of bcachefs that broken the kernel build, because in conflicts with others commits. But this can happen: it is not strictly related to the quality of code of bcachefs, but more to the interaction with other developers, that up to date it is the weak point of the project.

OpenDistro Leap/Tumbleweed/Aeon/…

OpenDistro should identify the community tools used for building the various distro. One distinctive aspect of OpenSUSE is how it is easy to create new distro, or custom repositories. So it should be the focus of the new naming scheme.