[No QA] Make GitHub Actions build asynchronously

[roryabraham]

Details

...Just something I worked on during a flight, mostly to get some much-needed practice with bash scripting. But still a nice time-saver!

This PR makes the GitHub Actions build script asynchronous, and it's considerably faster now.

Fixed Issues

n/a

Tests

Testing this was a pretty interesting task, and I mostly wanted to test two things:

1. Does it produce the same output?
2. Does it actually run faster?

To verify that the output is the same, I did the following:

1. Starting from an up-to-date main branch, create a test branch: git checkout temp-test
2. Make some changes to .github/libs/GithubUtils.js. This file is used by many GHA, so it's a good way to test the build script.
3. Run npm run gh-actions-build, then commit the changes.
4. git checkout main && git checkout Rory-AsyncGHActionsBuilds
5. Make the same changes to .github/libs/GithubUtils.js.
6. Run npm run gh-actions-build again
7. Run git diff temp-test... There should be no diff.
8. Cleanup by deleting your test branch: git checkout main && git branch -D temp-test

To verify that the script actually runs the build faster:

1. Add the following function to your bash profile (or just copy/paste it into your shell):

   # This is a basic shell utility to benchmark a bash command
   # @param $1 – The command to run
   # @param $2 - The number of time to run the command for benchmarking purposes
   function benchmark(){
     if [ -z "$1" ]; then
       echo "[Benchmark Err] – No command input"
       exit 1
     fi
     
     if [ -z "$2" ]; then
       echo "[Benchmark Err] – No number input"
       exit 1
     fi
     
     if [ "$2" -le 0 ]; then
       echo "[Benchmark Err] – Invalid number of uses – must be greater than 0"
       exit 1
     fi
     
     time eval "
         for ((i = 0; i < $2; i++)); do
           $1 &>/dev/null
         done
     "
   }

2. Then from an up-to-date main, run:

   > benchmark ./.github/scripts/buildActions.sh 1 
   
   real    0m15.119s
   user    0m20.259s
   sys     0m1.898s
   
   > benchmark ./.github/scripts/buildActions.sh 10
   
   real    2m38.679s
   user    3m32.631s
   sys     0m20.456s

3. Then on Rory-AsyncGHActionsBuilds run:

   > benchmark ./.github/scripts/buildActions.sh 1
   
   real    0m5.416s
   user    0m26.921s
   sys     0m2.615s
   
   > benchmark ./.github/scripts/buildActions.sh 10
   
   real    0m53.952s
   user    4m24.140s
   sys     0m24.770s

It's not a perfect experiment, but it does appear to be definitively faster – roughly 1/3 the execution time of a synchronous build.

Tested On

n/a – bash only

[deetergp]

Just a heads up git diff temp-test.. will compare what's committed to your feature branch against what's committed in temp-test. Simply doing git diff temp-test will compare whatever changes I made in your feature branch to what's committed to temp-test, which is probably what you meant, since your instructions did not expressly say to commit the changes to your feature branch after running npm run gh-actions-build

[deetergp]

NAB: TIL that ${#GITHUB_ACTIONS[@]} and ${#GITHUB_ACTIONS[*]} will return the number of elements in the GITHUB_ACTIONS array. Very cool!

[deetergp]

NAB: I've used !$ before, but $! (returning the PID of the last command) is a new one on me.

[deetergp]

Took me a minute to figure out what's going on, but it tests well and the code makes sense.

[OSBotify]

✋ This PR was not deployed to staging yet because QA is ongoing. It will be automatically deployed to staging after the next production release.

[OSBotify]

🚀 Deployed to staging in version: 1.0.73-4🚀

platform	result
🤖 android 🤖	success ✅
🖥 desktop 🖥	success ✅
🍎 iOS 🍎	success ✅
🕸 web 🕸	success ✅

[OSBotify]

🚀 Deployed to production in version: 1.0.74-0🚀

platform	result
🤖 android 🤖	success ✅
🖥 desktop 🖥	success ✅
🍎 iOS 🍎	success ✅
🕸 web 🕸	success ✅