oe-core needs to only get the minimal plumbing/capabilities, since there is no one size fits all solution to this, and many of the existing tools are far to heavy / complex for oe-core to carry (20-30 significant native and target depends/rdepends) .. this is due to broad (and deep) support in the tools, but is much more than is required for generating container images. In fact, what oe-core needs is somewhat custom, since the existing tools are not designed to quickly create simple OCI “image”. What is done with the image after it is generated (installed to a rootfs/image), pushed to a registry, converted to a docker compatible image, are secondary steps and are left as user specific extensions. The tools to support these manipulations are part of meta-virtualization.
Implementation is done. It is an open question whether or not this should go into oe-core this release, or soak for one release in meta-virt, before a proposed move to core.
Any decision on whether to move this from meta-virt to oe-core for 2.8?
There's enough other changes going into the fall release, that this is just too much. It needs more discussion and some broader meetings, which I'll do for the 2020 spring release. I'm moving this to 3.1 to show that, since IIRC, isn't 2.8 now 3.0 ?
Thanks Bruce. Yes, 2.8 now known as 3.0.
moving these to the next release. The ground work is in meta-virt, but obviously not ready for this release.
Still valid, moving to 4.2 so it can be coordinated with binary distro efforts.
Bulk move as requesting during YP BB meeting.
meta-virt has everything needed to do this now.