[peachbits]

Description

This branch speeds up npm test, which the precommit hook also runs. Each Jest test file starts with a fresh module registry, so anything jestSetup.js or a common import pulls in loads again in every one of the 141 files. Two suites also slept on real timers.

	develop	this branch
npm test, wall time	12.5s	7.1s
CPU time across all workers	95.3s	84.8s

These are medians of 5 alternating runs on a 16-core Mac, with a load average of 60 to 90 from other work. Every branch run passed all 141 suites and 1,393 tests. One develop run lost a Jest worker to a segfault in GiftCardActions.test.ts, which this branch doesn't touch.

1. Limit jest roots to src. Without watchman, Jest's haste map crawls every file under the project root on each run, including about 140k in node_modules. Every test and manual mock lives under src, so roots now points there.
2. Stop loading reanimated in every jest file. jestSetup.js called Reanimated's setUpTests(), which only registers the toHaveAnimatedStyle and toHaveAnimatedProps matchers. No test uses them, but the call loaded about 300 Reanimated modules into every test file, including the ones that never render a component.
3. Import date-fns format by path. date-fns 2 ships only CommonJS, so import { format } from 'date-fns' in intl.ts loaded all ~280 of its functions into every test file. It was the only import from the package root. The precommit hook took intl.ts off the ESLint warnings list.
4. Mock name lookups in the recipient network test. Picking Zcash through "Myself" runs a reverse name lookup against the live ZcashNames indexer. The reply took about 600ms and the test polls for about 1s, so a slow network failed it at random.
5. Use fake timers in the staking and rates tests. The staking stall tests waited out the provider's 2 second stall timeout and a 2.5 second slow node, and the bulk rates test slept 3 seconds. The two suites drop from 7.6s and 5.5s to 2.3s and 2.6s.

Testing:

• verify-repo.sh passes: install, prepare, ESLint on the changed files, and Jest.
• tsc passes.
• Both staking stall tests fail when the provider stops asking another node on a stall. The late-answer test fails right away with a diff when a stalled node's late answer takes over.

CHANGELOG

Does this branch warrant an entry to the CHANGELOG?

• Yes
• No

Dependencies

none

Requirements

If you have made any visual changes to the GUI. Make sure you have:

• Tested on iOS device
• Tested on Android device
• Tested on small-screen device (iPod Touch)
• Tested on large-screen device (tablet)