fragmented config
the current system with simple bash scripts (setup.sh, run.sh) is becoming unwieldy to maintain (and imho preventing us from supporting multiple platforms)
everything is being defined over and over. here's a few examples
we have $BUNDLER_PATH
's value set in
.gitlab-ci.yml
run.sh
setup.sh
likewise for other variables too
similarly, the Dockerfile mirrors the CI's before_script
. so changes to one, should also be reflected in the other
repetition like this is a recipe for misses (recent one was !97 (merged)). but it's plainly not simple to keep track of it all the time!
and to add to this, in #66 we (coughs i) see a need for conditionals to suppport CI pipelines in forks
another grievance i've had is that setup.sh and run.sh don't exit on error set -e
. this means the script finishes executing even with missing depenencies
particularly in a system without rsync
the prompts tend to mislead
✓ moved to assets/3rd-party/ folder
but setting -e
isn't entirely possible due to enforcing bundler version (0f2762e9) tending to exit with error at times (3dfaa241)
so more conditionals are needed, more complicated, file getting too large aaahhh! and changing one means changing in a lot of places too!