• [gentoo-dev] armbian-sources and ebuild-autogen announcement

    From Cor Legemaat@21:1/5 to All on Mon Jul 28 12:50:01 2025
    Hi:

    I am back after using Funtoo for a long time. For Funtoo after running
    into resource constrains, Daniel Robbins created an autogen system by
    witch the latest versions of packages get fetch thru scripts on a cron
    job. I wanted to get Gentoo running on embedded ARM boards but the only reliable boot chain I found was in Armbian. Funtoo's autogen is done in
    python but me and python != friends, so this started as a bunch of
    guile scripts to pull the patches used by Armbian and generate ebuilds
    from that in order to get updates from a cron job.

    The ebuild's are in my overlay at https://www.cor.za.net/code/cor/portage-overlay and auto generated from https://www.cor.za.net/code/cor/portage-overlay-src once a week.
    Installing virtual/armbian with the correct use flags for the cpu
    should install the required kernel, u-boot and arm-trusted-firmware to
    get going. The instructions to install the boot files are elog
    messages.

    Most packages in Funtoo where auto generated and running the latest
    versions. It surprised me how stable this was and that most updates do
    not require changes to the ebuild.

    https://www.cor.za.net/code/cor/ebuild-autogen is a package to model
    Funtoo's "doit" command to pul the latest updates for the package
    definitions and generate the ebuilds for them. 
    Differences from Funtoo:
    * The package definitions are in a git submodule in the autogen folder
    of the repository and not mixed with the output.
    * Putting ebuilds in the autogen definitions will be used as is in the
    output but if different from the one already in the output as a new
    release. So that key-worded or testing releases get auto generated
    and stable releases managed by fixed ebuilds.

    This is still alpha with samples in my repo but no documentation yet. I
    assume most open source projects have limited human resources so I want
    to propose this as a testing ground or discussion starter. To figure
    out what is needed in order to get unstable/testing releases into
    portage without human intervention. If a database is created with
    mappings between the packages in portage and the requirements in the
    package's build system like pypi pkg names, those could probably also
    be generated automatically with less mistakes than when done manually.

    Regards:
    Cor

    -----BEGIN PGP SIGNATURE-----

    iQIzBAABCgAdFiEEF21/f66PFCE7Ef8XtVLd72u7vq4FAmiHU4EACgkQtVLd72u7 vq4p2g//ft0uWyNIjg/34V1/C1M4rLWAOSk+kxJcMmzNJHT2Y/DsoCaiXkw2eO2b z/G6UyhaNkZWqMcGPG56wA8dW4Ruh7wDUY5bYlB08XbM5ZaLHtZPcgd/AabkcSIw Etz45iDsQcHvjbEdszKfoia8cdsVElvYgSPfpy/56tL1OXgUT5Mvvfs3b6Byvhzt mBSXmeunWPPYvK/KC98ZkdP0kBYPt1saaGAjJH+ulixRfapzAlhV/GDnwsrCiXXh q8sQe2sRUhjfkwi4GQUbFy0fCDRGgjX+aeZ6vlUv1UY0lAabZBRsb/QDf1zib1yF 9e2X3NpeunQ99ILbJaKNVf+UcyW/bqPoHK8j25z9/dJEv0n+43mVx3m4uBByMbdN J/uXoQD5xzahCBbuGKNob/tD3o6R0ZTNKhXzqUMq7Hxsu5wS8BoRP8JDKcmTK//4 6b6/Ylg2wZCT1Sr2Rq5R9GqHiL3CY3vjlLzE22prvF8YoQM172bQeqDnjXsGVcbx 12wjQ1AO7z7uzazROYFrdIYiIGkZae7WeO3CEuN+YJ7UqKuew5wiPGfMTqoFGmTA PVT/GjKrnK6XmluZnVAydAafV9wJQjE0P152VT0x9T38MZ7if6nMOUyEyBTlNNZs KEGkRcmOg9PKuCvUSjwba5mODh0F9ph9gVv0BtKsQtNvMyE0CSc=
    =Wuv6
    -----END PGP SIGNATURE-----

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)