akdev we missed a couple issues in my
@akdev we missed a couple issues in my PR
https://github.com/ublue-os/main/pull/310 I feel like thats a pattern for me... do something pretty good and working with good testing the first time... ⭐ Try to improve it a bit more... don't test quite as well... merge it... it's broken 😢 Thankfully in this case it didn't break the build, just didn't install the package as expected.
https://github.com/ublue-os/main/pull/310 I feel like thats a pattern for me... do something pretty good and working with good testing the first time... ⭐ Try to improve it a bit more... don't test quite as well... merge it... it's broken 😢 Thankfully in this case it didn't break the build, just didn't install the package as expected.
9 Replies
I'm TESTING a corrective PR now.
And to be specific... i had tested the jq script changes in the last PR... but not done a full image build...
I guess I got impatient.
GitHub
fix: use raw output from jq to install cosign again by bsherman · P...
The quotes included from the jq string did not behave as normal quotes when used in a bash variable, which resulted in an invalid url string for rpm-ostree install.
Ah
I’ve made this mistake many times
yeah, me too
I am a little surprised that the build.sh didn't error out
ahh, it's because the github-install script doesn't have strict bash settings, i'm going to fix that in this too
i'll add
set -ouex pipefail👍
I wish we could create an inventory of all the scripts we have and just put them somewhere centralize and run shellcheck on them
well, we could... just have a layer in an scratch image... or a github release tarball we extract on each build... but most of the
build.sh files are bespoke...
but we could do a lot of that in config repo...Maybe some shellcheck action that just finds script and runs shellcheck in all?
Then we could add it to the different parts
that's probably more appropriate
yeah
add it to config, akmods, main, nvidia
It's looking good now.
Approved