Implemented CI/CD

Implemented a CI/CD pipeline using GIthub Actions and improved the development workflow. 

Implemented workflows for automated linting, creating a release with auto changelog, and build compilation on multiple platforms plus code signing. 

Reduced the manual effort that could easily take between 1-2 hours down to 14 minutes.

Github Actions are really awesome.

Recently I was tasked with creating CI/CD that would add Android APK to releases in GitHub with every PR to main from an ionic codebase. I had never worked with ionic before and the person working on the codebase wasn’t that cooperative so I decided to dive in and explore the codebase to accomplish my task. Here are the steps I followed.

Step 1.
I created a bash script with all the initial plugins installation for android in the ionic codebase as shown in the image below.

Step 2.

1. I installed Cordova globally in the execution context of the workflow: npm i -g cordova
2. Installed dependencies with: npm i

Step 3.
Since I wasn’t using Android Studio in the workflow, I had to use jetifier so I installed it and run `npx jetify` to migrate Java, XML, POM, and ProGuard references that point to android.support.* packages, changing them so they point to the corresponding androidx.* packages.(As per docs on jetifier)

Step 4.
I was going to use capacitor to create my android build folder from which I would build the APK so I first built the source codes that generate a .www folder that capacitor uses to generate the android artifacts and then ran: 
`npx ionic capacitor copy android && npx cap sync android`

Step 5.
Move into the android folder and use gradlew to build the final APK with: `cd android && ./gradlew assembleDebug`
NB: For some reason, it doesn’t work without being in the android folder.

The final APK will be in `app/build/outputs/apk/debug/`

I have been exploring deeper automation with GitHub Actions and have written some custom self-hosted runners. Needless to say, I am a huge fun.

Set up CI / CD with Circle CI at work yesterday. I usually use GitHub Actions in my personal projects, but Circle CI’s ability to debug via SSH really won me over. I’m also super surprised by how fast it is! Medium containers start up incredibly fast, caches download quickly, and our lint / type checking pipeline finishes in less than 30 secs from front to back.

Also set up secrets management with Doppler. We had our secrets spread across several different services before, which made onboarding new developers more tedious. We had to add users to three or four different projects just for them to access the secrets locally. 

Doppler fixes that for us. It’s a single account and our developers have access to all the secrets they need. The CLI is super simple to use and the secrets are always up-to-date. It also syncs up with all of our deployment mechanisms via their API.

I learned a ton while working with the Deep Pack at TheTake as a software engineering intern. While a majority of the information about my job is proprietary and cannot be discussed with the public, there are a few things I can tell you:

• I analyzed the massive internal database kept by the company for common trends in the data and presented the information to the team in the form of various reports, a first of its kind analysis done by the company.
• I used this data to construct smaller datasets used to benchmark the performance of text search algorithms used in TheTake API.
• I wrote integration tests used in the company's CI/CD pipeline to track the performance of modifications to the text search algorithms.
• I proposed future optimizations to the text search algorithm.

This was my first internship ever. It was eye-opening to see how things are done at the professional level in software engineering. Working at a startup allowed me full access of every part of the stack, which helped immensely in my learning and skill building. I had amazing time working at TheTake and I wish them all the best in their future endeavors. 

Company website: https://thetake.ai/

Finishing up Screenshot testing Proof-of-Concept with Shot, Jetpack Compose and GitHub Actions.

Migrated a few views to their Jetpack Compose equivalents and started writing a couple of tests for these to see if we can detect changes and generate a report on both successes and failures. Also very helpful to see how much work it would be to rewrite part of the code and start using Jetpack Compose instead.