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
- clone this repo
- run
yarn install cd ios- run
pod install - run
gem install(?) - run
bundle exec fastlane ios cached_buildfor first time - check build time, and notice existens of folder
cached_derived_data - run
bundle exec fastlane ios cached_buildagain - 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 |
+------+------------------+-------------+
On CI, you can do the followwing in your build script:
- first try to download
<md5 of Podfile.lock>.zipfrom your favorite storage system (Azure Storage, S3 bucket, FTP, etc...) - If success, unzip into
ios/cached_derived_data - Run
bundle exec fastlane ios cached_build - If
<md5 of Podfile.lock>.zipdidn't exist, zipcached_derived_dataand upload it somewhere as<md5 of Podfile.lock>.zip
This way, only when Podfile.lock changes, Pods will be compiled.