Spike: investigate effort of switching from polkadot-js/api to papi

[shannonwells]

polkadot-js/api has given us (and others) problems, and papi was intended to address some of those. Investigate whether papi will suit our purposes and how difficult it would be to switch to papi.

[emresurmeli]

Migration Scope

• 49+ files importing @polkadot/api or @polkadot/types
• 3 core blockchain service files (~1,340 lines total) polkadot-api.service.ts, blockchain-rpc-query.service.ts, blockchain.service.ts
• Multiple workers (account, graph, content-publishing) using transaction submission
• Custom Frequency chain types via @frequency-chain/api-augment

Changes Required

1. Type System Overhaul - PAPI uses generated types from metadata, eliminating manual SCALE codec handling (.toNumber(), .toString(), etc.)
2. Transaction Submission - Fundamentally different API; affects all publisher/processor services
3. RPC/Query Methods - 50+ methods in blockchain-rpc-query.service.ts need refactoring
4. Chain Augmentation - Need to generate PAPI descriptors for Frequency; @frequency-chain/api-augment compatibility unknown
5. Testing - Complete mock/test infrastructure rebuild required

Key Risks

• Custom Frequency RPC methods may need special handling
• Redis-based nonce management may require rework
• Breaking changes across the entire blockchain interaction layer
• No clear migration path for @frequency-chain/api-augment
• PAPI ecosystem maturity for custom chains

Benefits

• ~50kB bundle vs ~200kB+ (lighter weight)
• Superior TypeScript support with auto-generated types from chain metadata
• Light client first architecture - better browser support
• Better performance - faster startup, less overhead
• Future-proof - recommended path forward by Parity

Thoughts

• Does the PAPI ecosystem need to mature further for custom chains like ours?
• Check for any examples emerging for similar migrations in the community

[shannonwells]

Meeting notes: Let's think about writing a parallel, papi-based blockchain rpc query service and blockchain service, and migrate one of the apps to it, and see how that goes. We'd have to decide which one to try first.