Reduce React Native iOS build times

Introduction

Amount of code in pods can be huge. Pods don’t change often. On CI, all pods are compiled over and over again, which is very time intensive. What if we compile Pods once and use the result over and over again? This repo contains a demo how to accomplish this.

Based on this blogpost

Demo

• clone this repo
• run yarn install
• cd ios
• run pod install
• run gem install (?)
• run bundle exec fastlane ios cached_build for first time
• check build time, and notice existens of folder cached_derived_data
• run bundle exec fastlane ios cached_build again
• check build time

If everything worked well, you'see a drastically smaller build time the second time.

First build, without cache:

+------+------------------+-------------+
|           fastlane summary            |
+------+------------------+-------------+
| Step | Action           | Time (in s) |
+------+------------------+-------------+
| 1    | default_platform | 0           |
| 2    | gym              | 384         |
+------+------------------+-------------+

Second build:

+------+------------------+-------------+
|           fastlane summary            |
+------+------------------+-------------+
| Step | Action           | Time (in s) |
+------+------------------+-------------+
| 1    | default_platform | 0           |
| 2    | gym              | 63          |
+------+------------------+-------------+

CI

On CI, you can do the followwing in your build script:

1. first try to download <md5 of Podfile.lock>.zip from your favorite storage system (Azure Storage, S3 bucket, FTP, etc...)
2. If success, unzip into ios/cached_derived_data
3. Run bundle exec fastlane ios cached_build
4. If <md5 of Podfile.lock>.zip didn't exist, zip cached_derived_data and upload it somewhere as <md5 of Podfile.lock>.zip

This way, only when Podfile.lock changes, Pods will be compiled.

Presentation

https://docs.google.com/presentation/d/e/2PACX-1vTfrXqbjTRAYCABpStRknZeCX83ku_MOzGsa0ZOq7D_JPqcYxBrrc67hLOsK1RQnKV7i94zJnAmPubZ/pub?start=false&loop=true&delayms=3000