Nightly Builds

    hello ( heitbaum)

    it seems to be "Something is rotten in the state of Denmark" again
    for /12.0/RPi/RPi5/ is being built
    LibreELEC-RPi5.aarch64-12.0-nightly-20240819-7e0b15d.img.gz (last)

    and for /13.0/RPi/RPi5/ not
    LibreELEC-RPi5.aarch64-13.0-nightly-20240816-018c547.img.gz (last)

    Display Spoiler

    Changelog
    2024-08-19 (a25491a): #9202 ffmpegx: arm: allow 32-bit arm compile with binutils 2.43
    2024-08-18 (5e1c06f): #9201 kodi: update to cdfc755
    2024-08-18 (984e99a): #9200 gcc: 14.2 upstream fix for building libaom on arm64
    2024-08-17 (14951f7): #9198 linux (RPi): update to 6.6.45
    2024-08-17 (93b49f6): #9194 systemd: add ZTERCR21 BT remote device to 70-local-keyboard.hwdb
    2024-08-17 (58ae4a6): #9195 update_binary-addons: add -x option to skip no tag updates
    2024-08-17 (43f76dc): #9193 kodi-binary-addons: update to latest versions
    2024-08-16 (018c547): #9192 mesa: drop unnecessary kmsro option
    2024-08-16 (35a05c8): #9190 Package updates
    2024-08-16 (d0f92f4): #9186 proftpd: update addon (1), convert HTML special entities
    2024-08-16 (a05f732): #9103 mesa: update to 24.2.0
    2024-08-15 (f02cf5d): #9185 inputstream.ffmpegdirect: update 22.0.0-Piers to 22.0.1-Piers
    2024-08-15 (b0a8fed): #9184 Package updates

    What could be the reason for this (I have 0 idea here)?

    with best regards jd

    (info translated with DeepL)

    Your probably reading too much into it. The build system runs many workflows and then many jobs within those workflows. We have been doing extra LE12 builds recently which will push out or cancel other builds. When a workflow runs - it runs 2 jobs at a time (in random order for the target architectures) and will pick the hash at the time of the job kick off - not the workflow kickoff.

    Whilst the job is scheduled nightly - it doesn’t mean there will always be a nightly.


    You can always check the scheduler for LE13 builds at https://github.com/LibreELEC/acti…/build-LE13.yml
    Or in total at https://github.com/LibreELEC/actions/actions

    may help the crystal ball be less cloudy.

    Again, what change triggered this?

    Same build# on 2 builds with different date.

    LibreELEC-RPi5.aarch64-12.0-nightly-20240829-a7e7ad8
    LibreELEC-RPi5.aarch64-12.0-nightly-20240831-a7e7ad8

    I have explained it a few times now. But I will again for clarity.

    Jobs are scheduled (if there are changes in the last 24 hours and an image will be built, it does not check it an image of that hash exists already.)

    Jobs are also run manually, in this case the RPi ones were run to address a RPi 2GB built “quickly.”


    There is nothing to be read into the images with different dates and the same hash. There maybe many of the same hash, you are welcome to use any off them, though the contents will have different compile dates, they will be from the same source.


    Hope this fully answers the questions.