The Linux ecosystem just sent a message to Apple and other proprietary vendors: we're not asking permission to read your filesystems anymore. We're building it ourselves.
A new version of bcachefs 1.37.0 appeared as a next-generation copy-on-write snapshotting GPL filesystem for Linux, and it brings Linux 7.0 kernel code compatibility, with erasure coding moving from experimental to complete functionality, plus automatic recovery from devices with bad flush support and faster recovery from unclean shutdowns. This matters more than it might sound. The filesystem is no longer embedded in the Linux kernel itself, but it can now be loaded as a DKMS module, reflecting the messy reality of kernel maintenance and developer politics.
Here's the deeper story. When Linus Torvalds removed all bcachefs code from the Linux kernel beginning with version 6.18, the project didn't die. Instead, it proved that modern open-source infrastructure can sustain important work outside the monolithic kernel. Bcachefs 1.37 brings a major update to its Principles of Operation documentation, expanding it to around 100 pages, suggesting serious development effort is happening despite the separation.
Meanwhile, Apple's APFS filesystem has long been a thorn in Linux's side. Apple launched APFS a decade ago, and by default Linux still cannot mount or read APFS volumes. That's a real problem for users with Apple hardware who want to access their data on Linux, or anyone caught between ecosystems. Commercial vendors charge money for this privilege; the Linux community typically says no thank you and reverse engineers instead.
KDE Linux now has support for reading from and writing to disks formatted with Apple's APFS filesystem, though it's early days. Read-write support for APFS is available via a DKMS out-of-tree kernel module with KDE Linux installing the linux-apfs-rw-dkms package. The linux-apfs-rw out-of-tree module is still considered experimental, especially for write support. In plainer terms: it works in alpha, don't store critical data there yet.
The technical caveats matter. The APFS module is the result of reverse engineering with limited testing, and if using write support there is a real risk of data corruption, so mounts are read-only by default. This is honest software engineering; developers are telling users exactly what they're getting into rather than pretending it's production ready.
Both projects illustrate something important about how the open-source model actually works. When faced with proprietary or unsuitable solutions, the community doesn't accept vendor decisions as fixed. Bcachefs got kicked out of the kernel by Linus Torvalds over governance disputes, but the work continues. Apple maintains APFS as a closed ecosystem, but Linux developers spent years reverse engineering it anyway. These aren't victories in the traditional sense. They're acts of determined pragmatism in a world where you can't always get what you want, so you build it yourself.
The question isn't whether bcachefs and APFS support are perfect. They're not. The question is whether users should be locked into a single ecosystem or have options. On that measure, Linux just improved.