Recently several new projects came into being to work stand-alone that possibly, when ready, though some are working currently, can be used with B/LFS.

One is vdev found here:

https://github.com/jcnelson/vdev

which implements a new style of device management using events and a filesystem structure without the need for dbus. The author has posted on the Slackware forum about the project and is making efforts to add a backwards compatibility layer and wrapper library for libudev and libgudev possibly.

Next up is loginkit found here:

https://github.com/dimkr/LoginKit

which provides a stand-alone logind, pam-systemd, and libsystemd library shim kit without a full needless systemd implementation or specific init structure. The author has used it to wrap ConsoleKit2, from here: https://github.com/ConsoleKit2/ConsoleKit2 to achieve a modern session manager.

Lastly, although mentioned previously, the newest release has opened new doors for project portability, which is s6 found here:

http://skarnet.org/software/

which provides a new type of service supervision similar to Runit and DJB's daemontools, similar somewhat to systemd, but less intrusive and uses named pipes and FIFOs rather that resource eating UNIX sockets. The only shortfall is currently, the init-utilities lack a wrapper function so doing things actually like init, halt, poweroff, reboot, and shutdown must be called through other utilities unless wrapped via scripts.

(Runit-for-LFS has some wrapper utilities that can be functional tools mind you for s6, but we lack currently the essential init binary or script to boot a system with before passing off to s6. s6 does have a working example for stage-1 init but it may need some fix-ups to be usable. We then would have to write a stage-2 to do mounting filesystems, load modules, etc. as well as a stage-3 shutdown script, and possibly scripts for other functions.)

The alternative would be CLFS/CBLFS.

CLFS would be the alternative to LFS, not for it. These are alternatives to existing software packages.

I would be interested in how vdev progresses. I recently managed to build a BLFS system that used Busybox mdev, and that's when I was reminded of how many things in a BLFS system rely on udev/gudev. I was able to fix or work around most of those things satisfactorily, and the system was functional. But in the end, I think that issue is mostly why I didn't continue on with mdev.

We may have a breakthrough with s6 Arthur. I found out how the stage-1 init script is setup and fairly soon we may have a working model.