chore: trade apk size to speed up launch time

[siddarthkay]

Summary

This PR un compresses the android bundle which apparently improves Time to Interactivity by 400ms.
There is a slight tradeoff in bundle size and I want to measure that.

This technique was first discovered and accepted by RN team here : facebook/react-native#49449

Platforms

• Android

status: ready

[status-im-auto]

Jenkins Builds

❔	Commit	#️⃣	Finished (UTC)	Duration	Platform	Result
✔️	531f1ce	#1	2025-02-18 18:19:42	~4 min	tests	📄log
❌	531f1ce	#1	2025-02-18 18:23:39	~8 min	android-e2e	📄log
✔️	531f1ce	#1	2025-02-18 18:24:29	~9 min	android	🤖apk 📲
✔️	531f1ce	#1	2025-02-18 18:25:35	~10 min	ios	📱ipa 📲
✔️	19d202b	#2	2025-02-25 15:51:46	~5 min	tests	📄log
❌	19d202b	#2	2025-02-25 15:55:19	~8 min	android-e2e	📄log
❌	19d202b	#2	2025-02-25 15:55:41	~9 min	android	📄log
✔️	19d202b	#2	2025-02-25 15:57:23	~10 min	ios	📱ipa 📲

[ulisesmac]

Hey @siddarthkay !

Thaks for this PR, I'm wondering, what was the previous bundle size and what is the new one?

[siddarthkay]

There is an APK size increase of 11.7 MB

[ilmotta]

Nice finding @siddarthkay! The 400ms improvement can vary a lot per app it seems. This number comes from the Discord app from the linked RN PR right? If so, before merging this PR I think it's important we show evidence of sufficient improvement to justify the increase in bundle size. Let's say in 2-3 Android devices sufficiently different and averaging the test runs.

For this, it seems that a QA won't be able to test this PR because we need precise measurements.

[churik]

@siddarthkay @ilmotta If there are any insights on how to measure this parameter, I can allocate resources to help.

I tried checking it via Profiler in Android Studio but couldn’t find any relevant data—most likely, it needs to be built locally (or maybe I just don’t know where to look).

[siddarthkay]

@churik : the measurement was done using react-native-performance I can find you some notes on how to do so.
I had played with this tool a while ago.

@ilmotta : yes it was done on Expensify app too along with Discord.
I too will run some benchmarks locally.
They key metric this approach improves is TTI ( Time To Interactivity ) at the cost of increase in APK size due to an uncompressed bundle.

[ilmotta]

@ilmotta : yes it was done on Expensify app too along with Discord. I too will run some benchmarks locally. They key metric this approach improves is TTI ( Time To Interactivity ) at the cost of increase in APK size due to an uncompressed bundle.

Hoping for the best in your benchmarks @siddarthkay 🙏🏼 To judge the tradeoff is worth it we will need a statistically significant result. I volunteer to help if you can provide the steps, as it would be quite important to verify in a diverse set of devices.