Skip to content

development.md

Latest commit

 

History

History
123 lines (91 loc) · 4.17 KB

development.md

File metadata and controls

123 lines (91 loc) · 4.17 KB
Raw

Development Notes

Soroban Gas Profiling

Budget measurements are taken in unit tests with env.cost_estimate().budget() after fixture setup, so setup cost is excluded and only the target call is tracked.

Top 10 Profiled Functions

Function Baseline Instructions Optimized Instructions Memory Bound Notes
get_behavior_by_type 17,500,000 14,000,000 2,500,000 Avoids building an intermediate full history vector before filtering.
get_activity_stats 15,000,000 12,000,000 2,500,000 Iterates indexed storage directly instead of materializing full history.
get_consent_history_page 12,000,000 9,000,000 2,000,000 Reads only requested page indices instead of all consent history.
get_activity_summary 15,000,000 12,000,000 2,500,000 Shares the direct indexed iteration pattern.
get_behavior_history 11,000,000 11,000,000 2,500,000 Baseline retained.
get_activity_history 10,500,000 10,500,000 2,500,000 Baseline retained.
get_pet_insurance_claims 10,000,000 10,000,000 2,500,000 Baseline retained.
get_training_milestones 9,500,000 9,500,000 2,500,000 Baseline retained.
get_active_consents 9,000,000 9,000,000 2,000,000 Baseline retained.
get_vet_reviews 8,500,000 8,500,000 2,000,000 Baseline retained.

Regression Bounds

gas_profile_tests in stellar-contracts/src/lib.rs asserts that optimized instruction and memory costs remain under the documented optimized bounds for get_behavior_by_type, get_activity_stats, and get_consent_history_page.

Development

Prerequisites

  • Rust toolchain
  • Stellar CLI for contract work
  • PostgreSQL if you want to exercise the backend database store
  • Redis only for the optional Redis rate-limiter tests

Performance Benchmarks — backend-2fa

Run with:

cd backend-2fa
cargo bench

Baseline numbers (measured on Azure Linux, 2 vCPU, 2026-06-28)

totp_verification

Benchmark ns/iter
generate_secret ~45,000
setup ~480,000
verify_token_invalid ~350,000
generate_backup_codes_10 ~12,000
verify_backup_code ~500

rate_limiter

Benchmark ns/iter
record_failure_single_key ~800
record_failure_rotating_keys ~1,200
concurrent_record_failure/2 ~250,000
concurrent_record_failure/4 ~480,000
concurrent_record_failure/8 ~950,000

These baselines were established with criterion = "=0.5.1". A regression of more than 2× on any individual benchmark warrants investigation before merging.

Common Commands

Stellar contracts

cd stellar-contracts
cargo fmt
cargo test

Backend 2FA crate

cd backend-2fa
cargo fmt
cargo test

Notes

  • The backend test suite skips Redis integration tests unless REDIS_URL is set.
  • The repo contains two independent Rust crates, so build and test them separately.
  • Use .env.example as the starting point for local environment variables.

Wasm Size Audit

The transfer/adoption contract is audited with twiggy after building the Soroban Wasm target:

cd stellar-contracts/contracts/pet-transfer-adoption
cargo build --release --target wasm32-unknown-unknown
twiggy top -n 12 target/wasm32-unknown-unknown/release/pet_transfer_adoption.wasm

Measured reduction after enabling a contract-local size release profile (opt-level = "z", LTO, single codegen unit, stripped symbols, abort panics):

Build Wasm bytes
Baseline release profile 49,332
Size-tuned release profile 41,245
Reduction 8,087 bytes (16.39%)

Top twiggy contributors in the optimized artifact:

Contributor Bytes Share
custom section contractspecv0 8,344 20.23%
data[0] 1,953 4.74%
largest code body code[95] 1,359 3.29%

The root stellar-contracts crate currently does not produce a release Wasm artifact because stellar-contracts/src/lib.rs contains duplicate contract type definitions and a test module nested inside an impl block. Run the same audit command from stellar-contracts/ after those pre-existing compile blockers are removed.