I lost a few hours over the weekend because the GH release uploads
are flakey and the action I was using doesn't internally perform
retries. I had to manually delete the failed uploads from the release
and then re-trigger the builds across several platforms, several times
for both of the releases I pushed this weekend :-/
This commit speculatively switches over to using the GH cli in the
hopes that the error reporting is better, and also because it is
simpler to externally drive a retry loop.
Let's see how this goes.
This commit reduces the cron schedule for nightly builds down from
hourly and to actually nightly (at ~3am).
The push-to-master flow now triggers effectively the same thing at push
time, so if anything, this will reduce the latency of the continuous
package build and deploy.
The push-to-master flow will now skip builds if it changes only the
docs.
This persuades the CI to install both the arm and intel flavors
of the rust toolchain on macOS, and the deploy script to generate
a universal binary.
* need big sur to build for M1
* Use cross-compilation compatible mlua from my fork for now
The mac builds started to fail with a crate resolution issue that I
can't reproduce locally shortly after I updated the cache action. Let's
tweak the cache key to see if that clears things up.
tweak some container names
fixup container names and regenerate the config
Ensure that curl is installed
the rust toolchain stuff wants curl
run apt update before apt install
centralize updating apt, and install sudo
revise get-deps script for slimmer debian images
more container related tweaks
get-deps: Don't require lsb-release on debianish systems
more get-deps improvements
Explicitly ask GH actions to recursively checkout the repo
Explicitly install git
fetch tags and tweak git build for debian systems
moar adjustments
remove deb8 (its wayland is too old), fixup debian publish
Ensure git build deps are installed always, tidy up sudo usage
