Workflow for building custom packages?

[doragasu]

I am new to armbian, and so far I managed to build a working image with some custom packages added and some removed by customizing `userpatches/lib.config`. Easy stuff and it's working great so far!

 

Now I need to build some packages from sources and get them added to the resulting image. The usual way I would do this, would be installing a cross toolchain, having a staging dir with includes and libs from the dev packages I add to the image, etc. But after building the base armbian image, on my system I see I have two docker images related to armbian (ghcr.io/armbian/docker-armbian-build and armbian.local.only/armbian-build), and I wonder if maybe the "local only" image is somehow like a "staging" image or how does this work.

 

So far I have not been able to find documentation about the procedure, is there any guide explaining how to achieve this?

[Werner]

Easiest way would be using the customization script: https://docs.armbian.com/Developer-Guide_User-Configurations/#user-provided-image-customization-script

[doragasu]

Thanks for help. If I understand correctly, that script is useful to add files to the image and perform modifications of it. But it seems that runs on a virtualized environment (QEMU I presume). Although that might work, I would prefer having a cross-compilation environment. I have verified the docker images I mentioned before have the cross compile toolchain, but I don't know if armbian generates something like a staging directory to be used for cross compilation. Could it be this is not possible with armbian and I should be using something like Yocto for this task?