Implement tests as GitHub workflow

[findgriffin]

Add scripts to setup DB using local container and validate sample app:

All of this work should be copyable to the other sample apps (e.g. Python).

1. Setup GitHub Actions workflow in .github to install JVM, fauna docker container, and fauna-shell.
2. Use fauna shell in test/setup.sh to configure the test database.
3. Run curl requests against the sample app in test/validate.sh

[cynicaljoy]

We should check with the Fauna shell team to be sure no files are intended to be checked in when working on a team.

[cynicaljoy]

from the discussion yesterday it sounds like we want to produce a test suite that's more native to the sample app. So JUnit in this case.

[cynicaljoy]

if we're producing JUnit this doesn't matter much, but providing feedback for general bash improvements.

1. I don't think we should write files unless absolutely required
2. use a var for all of the common curl options, and don't use shorthand in scripts (we still have a lot of this everywhere, but trying to be better about it)
3. separate flags to new lines

result:

CURL_OPTS='--silent --header "Accept: application/json" --header "Content-Type: application/json"'

PAGE_ONE_=$(curl $CURL_OPTS \
    --retry-all-errors \
    --connect-timeout 5 \
    --max-time 10 \
    --retry 5 \
    --retry-delay 10 \
    --retry-max-time 60 \
    "$ENDPOINT/products?pageSize=1")

[findgriffin]

I agree with all 3 points here.

On points 1 and 2, I tried to use the syntax you proposed here, but it doesn't seem to work. CURL_OPTS is not getting split into multiple arguments, and it seems to hang when I try and pass the output of curl into a variable. 🤔

[cynicaljoy]

you could eval, but I actually think this pattern would be best:

CURL_OPTS=(
  "--silent"
  "--show-error"
  "--fail"
  "--connect-timeout" "2"
  "--max-time" "2"
)

# just throwing in a verbose to demonstrate we can append more opts if needed
curl "${CURL_OPTS[@]}" --verbose "$ENDPOINT/products?pageSize=1"