Critical: Error Handling and Code Quality Refactoring Needed

[codewizdave]

Error Handling and Code Quality Refactoring

Priority: Critical
Status: Discussion Required
Type: Architecture / Code Quality
Created: 2025-02-04
Tags: architecture, error-handling, functional-programming, ipc, database, hooks

---

Problem Statement

The codebase has accumulated significant technical debt related to error handling and code quality. Several architectural issues make the application difficult to maintain, debug, and extend:

Current Issues

1. Unsafe Navigation Patterns

• Location: src/routes/locations.tsx, src/routes/zones.tsx, src/routes/sectors.tsx
• Problem: Using useEffect for navigation is an anti-pattern in React
• Impact: Can cause infinite loops, race conditions, and violates React best practices
• Example:

useEffect(() => {
  if (warehouses?.length === 0) {
    navigate({ to: "/onboarding/welcome" })
  }
}, [warehouses, navigate]) // navigate changes on every render

2. Confusing Data Fetching Logic

• Location: All data fetching routes
• Problems:
  • useWarehouses() is called but its result is often ignored
  • useLocations(undefined) is passed but intent is unclear (get all? get first?)
  • Fallback logic inside hooks (warehouseId || firstWarehouse.id) creates confusion
• Impact: Unpredictable behavior, hard to understand what data will be fetched

3. "God Hook" Anti-Pattern

• Location: useBackend() hook
• Problems:
  • Single hook handles all IPC communication
  • No separation of concerns
  • Difficult to test and mock
  • Violates Single Responsibility Principle
• Impact: Tight coupling, low reusability, difficult to extend

4. Broken Type Chain with Global Access

• Location: Throughout the codebase where Electron IPC is accessed
• Problems:
  • Using (window as any).electronAPI breaks the entire type chain
  • Creates tight 3D coupling between renderer and main process
  • No type safety for IPC calls
  • Impossible to refactor safely
  • Global mutable state
• Impact: Type errors at runtime, impossible to track IPC usage, unsafe refactoring

5. Inadequate Error Handling

• Location: safeIpcCall() and throughout the codebase
• Problems:
  • safeIpcCall() returns fallback values but errors are silently ignored
  • No structured error types
  • No retry logic for transient failures
  • No user feedback for errors
  • No logging/monitoring
• Impact: Errors are swallowed, impossible to debug, poor UX

---

Backend Issues (src/backend/)

6. Fake Service Layer Anti-Pattern

• Location: src/backend/services/*.ts
• Problems:
  • *-service.ts files exist but are just thin wrappers
  • No actual business logic, just pass-through functions
  • Adds unnecessary indirection
  • Violates YAGNI principle
• Example:

// services/analysis-service.ts
export const performABCAnalysis = (...args) => {
  return runABCAnalysis(...args)  // Useless wrapper!
}

• Impact: Code bloat, false sense of architecture, harder to navigate

7. Global Mutable State - Plugin Registry

• Location: src/backend/import/plugins/registry.ts
• Problems:
  • export const registry: PluginRegistry = {} is global and mutable
  • registerPlugin() mutates global state at runtime
  • No way to know which plugins are registered without side effects
  • Testing nightmare (shared state between tests)
• Impact: Race conditions, unpredictable behavior, impossible to test in isolation

8. Runtime require() Instead of Static Imports

• Location: src/backend/analysis/abc-analysis.ts:33, dead-stock-analysis.ts:21
• Problems:
  • const { getProductMovementTotals } = require('../database/queries')
  • Hides dependencies from compiler
  • No type checking at import time
  • Impossible to tree-shake
  • Errors only appear at runtime
• Impact: Lost type safety, hidden dependencies, production crashes

9. Dynamic Imports in Electron Main Process

• Location: src/backend/services/import-service.ts:109, 225
• Problems:
  • const { loadToDatabase } = await import('../import/loader')
  • No code splitting benefit in Electron main process
  • Hides dependencies
  • Makes code tracing difficult
• Impact: No performance gain, added complexity

10. No Data Validation

• Location: src/backend/import/loader.ts
• Problems:
  • All insert* functions accept readonly any[]
  • No Zod validation or schema checking
  • Malformed data silently fails or crashes
• Example:

export const insertWarehouses = (warehouses: readonly any[]): number => {
  return bulkInsert(warehousesTable, warehouses, (w) => ({
    // No validation that w has correct fields!
  }), 'Warehouses')
}

• Impact: Data corruption, cryptic errors, impossible to debug

11. Massive Code Duplication

• Location: src/backend/import/loader.ts
• Problems:
  • Same pattern repeated 20+ times for each entity
  • 750+ lines doing essentially the same thing
  • Every entity has identical insert* function
• Example:

export const insertWarehouses = (warehouses: readonly any[]): number => {
  return bulkInsert(warehousesTable, warehouses, (w) => ({...}), 'Warehouses')
}
export const insertUsers = (users: readonly any[]): number => {
  return bulkInsert(usersTable, users, (u) => ({...}), 'Users')
}
// ... 18 more times

• Impact: Maintenance nightmare, error-prone, violates DRY

12. God Function - loadToDatabase

• Location: src/backend/import/loader.ts:621-752
• Problems:
  • 131 lines in a single function
  • 20+ if blocks checking each array
  • No abstraction, just procedural code
• Impact: Untestable, unreadable, unmaintainable

13. TODOs in Production Code

• Location: src/backend/import/mapping-service.ts:282, 292
• Problems:
  • saveMappingPreset() and loadMappingPreset() are stubs
  • Functions do nothing but log
  • No implementation of persistence
• Impact: Features appear to work but don't actually persist data

14. Type Safety Lost with SQL Queries

• Location: src/backend/database/index.ts:177-183
• Problems:
  • Raw SQL queries return any[]
  • All type information lost at database boundary
  • prepare('SELECT * FROM warehouses ORDER BY id').all() returns any
• Example:

export const getAllWarehouses = () => {
  const db = getDbRaw()
  return db
    .prepare('SELECT * FROM warehouses ORDER BY id')
    .all()  // Returns any[], no typing!
}

• Impact: Complete loss of type safety, runtime type errors

15. Global Singleton with Mutable State

• Location: src/backend/database/index.ts:15
• Problems:
  • let sqliteDb: any | null = null (global mutable, plus any!)
  • Singleton pattern makes testing impossible
  • No way to reset or mock database in tests
  • any type removes all type safety
• Impact: Untestable code, shared state, lost type safety

16. Silent Error Handling

• Location: src/backend/import/loader.ts:44
• Problems:
  • Try/catch logs errors but continues execution
  • No way to know if insertions failed
  • No Result types, just console.error
• Example:

try {
  db.insert(table).values(transform(item)).run()
  inserted++
} catch (error) {
  console.error(`Error inserting ${entityName}:`, error)
  // Continues anyway, no way to track failures!
}

• Impact: Silent data loss, impossible to debug

17. Hard-coded String Literals

• Location: Throughout backend
• Problems:
  • Status types as strings: 'success' | 'failed'
  • Movement types: 'inbound' | 'outbound' | 'transfer'
  • No type safety, typos possible
  • Refactor impossible
• Impact: Runtime errors from typos, impossible to refactor safely

18. Production Logging with console.log

• Location: Throughout backend
• Problems:
  • console.log('🎲 [MOCK DATA] Starting generation...')
  • console.log('✅ [DB INSERT] Warehouses:', ...)
  • No structured logging
  • Impossible to disable in production
  • Performance impact
  • No observability
• Impact: Performance degradation, no monitoring, cluttered code

19. Inconsistent Naming Conventions

• Location: Throughout backend
• Problems:
  • SCHEMAS vs ipcContract vs registry
  • performABCAnalysis vs runDeadStockAnalysis
  • warehouseId vs plugin_id (camelCase vs snake_case)
• Impact: Confusion, cognitive load, errors

---

Proposed Solution

Functional Programming Approach

We should adopt a functional programming approach with the following patterns:

1. Result Type Pattern

Replace throwing exceptions with explicit error handling using discriminated unions:

// Discriminated union for explicit error handling
type Result<T, E = AppError> =
  | { readonly success: true; readonly data: T }
  | { readonly success: false; readonly error: E }

// Helper functions (pure, composable)
const success = <T>(data: T): Result<T> => ({ success: true, data })
const failure = <E>(error: E): Result<never, E> => ({ success: false, error })

// Transformations
const map = <T, U, E>(
  result: Result<T, E>,
  fn: (data: T) => U
): Result<U, E> =>
  result.success ? success(fn(result.data)) : result

const chain = <T, U, E>(
  result: Result<T, E>,
  fn: (data: T) => Result<U, E>
): Result<U, E> =>
  result.success ? fn(result.data) : result

// Benefits:
// - Type-safe: forced to handle errors explicitly
// - Composable: map, chain, flatMap operations
// - Immutable: readonly properties prevent mutations
// - No exceptions: predictable control flow

2. Union Types for Error Domains

Use union types instead of enums for better type safety and functional programming:

// Union type for error domains
type ErrorDomain =
  | 'DATABASE'
  | 'IPC'
  | 'VALIDATION'
  | 'NETWORK'
  | 'BUSINESS'

// Error type as a product type (readonly for immutability)
type AppError = Readonly<{
  domain: ErrorDomain
  code: string
  message: string
  cause?: unknown
  context?: Readonly<Record<string, unknown>>
  timestamp: Date
  recoverable: boolean
}>

// Constructor functions (pure)
const createError = (
  domain: ErrorDomain,
  code: string,
  message: string,
  options?: {
    cause?: unknown
    context?: Record<string, unknown>
    recoverable?: boolean
  }
): AppError => ({
  domain,
  code,
  message,
  cause: options?.cause,
  context: options?.context,
  timestamp: new Date(),
  recoverable: options?.recoverable ?? false
})

// Specific error creators
const databaseError = (
  operation: string,
  table: string,
  cause: unknown
): AppError =>
  createError('DATABASE', `DB_${operation.toUpperCase()}_FAILED`,
    `Failed to ${operation} on ${table}`,
    { cause, context: { table, operation } }
  )

const ipcError = (
  channel: string,
  cause: unknown
): AppError =>
  createError('IPC', 'IPC_CALL_FAILED',
    `IPC call to '${channel}' failed`,
    { cause, context: { channel }, recoverable: true }
  )

const validationError = (
  field: string,
  value: unknown,
  message: string
): AppError =>
  createError('VALIDATION', 'VALIDATION_FAILED',
    message,
    { context: { field, value } }
  )

3. IPC Wrapper with Retry Logic

Create a wrapper that handles:

• Timeout management
• Automatic retry with exponential backoff
• Structured error logging
• Proper error propagation

const safeIpcCall = async <T>(
  channel: string,
  data?: unknown,
  options?: IpcOptions
): Promise<Result<T, AppError>>

4. tRPC Pattern with Direct API Calls

No service layer - direct function calls through tRPC:

// Backend API - just functions, no classes, no services
type BackendApi = {
  readonly locations: {
    readonly getAll: (warehouseId?: string) => Promise<Result<LocationsData, AppError>>
    readonly getById: (id: string) => Promise<Result<Location, AppError>>
    readonly create: (data: CreateLocationDto) => Promise<Result<Location, AppError>>
  }
  readonly zones: {
    readonly getAll: (warehouseId?: string) => Promise<Result<ZonesData, AppError>>
    readonly getById: (id: string) => Promise<Result<Zone, AppError>>
  }
  // ... for all entities
}

// Usage through tRPC - direct calls, no intermediate services
const { data } = await api.locations.getAll()
// Returns Result<LocationsData, AppError>

// In React Query / tRPC hooks
const useAllLocations = (warehouseId?: string) => {
  return useQuery({
    queryKey: ['locations', warehouseId],
    queryFn: () => api.locations.getAll(warehouseId),
    // Result<T, E> automatically converted to throw on error
  })
}

// Benefits:
// - No service layer overhead
// - Direct function calls (composable, testable)
// - tRPC provides type safety end-to-end
// - Clear contract: functions take X and return Y

5. React Query Integration

Integrate with React Query properly:

• Convert Result<T, E> to throw errors for React Query's built-in error handling
• Use retry option based on error.recoverable flag
• Clear separation between data fetching and business logic

6. Navigation Pattern

Replace useEffect navigation with:

• Router-based redirects in loaders
• Conditional rendering
• Protected routes pattern

7. Type-Safe IPC Layer with Zod Contract

Replace (window as any).electronAPI with a Zod-based contract system:

// shared/ipc/contract.ts - Single source of truth
import { z } from 'zod'

const locationSchema = z.object({
  id: z.string(),
  code: z.string(),
  type: z.string(),
  // ... all fields
})

const locationsDataSchema = z.object({
  locations: z.array(locationSchema),
  kpis: z.object({
    totalLocations: z.number(),
    // ... all KPIs
  }),
})

// IPC Contract - defined once, shared between frontend/backend
const ipcContract = {
  locations: {
    getAll: {
      input: z.object({ warehouseId: z.string().optional() }).optional(),
      output: locationsDataSchema,
    },
  },
  zones: {
    getAll: {
      input: z.object({ warehouseId: z.string().optional() }).optional(),
      output: zonesDataSchema,
    },
  },
} as const

// Type inference from Zod schemas
type IpcInput<T extends keyof IpcContract, M extends keyof IpcContract[T]> =
  z.infer<IpcContract[T][M]['input']>
type IpcOutput<T extends keyof IpcContract, M extends keyof IpcContract[T]> =
  z.output<IpcContract[T][M]['output']>

// Type-safe proxy generator
type IpcProxy<T extends IpcContract> = {
  [K in keyof T]: {
    [M in keyof T[K]]: (
      input: IpcInput<K, M>
    ) => Promise<Result<IpcOutput<K, M>, AppError>>
  }
}

const createIpcProxy = <T extends IpcContract>(
  channels: unknown
): IpcProxy<T> => {
  // Recursive proxy that validates input/output with Zod
  return createProxy(channels, ipcContract)
}

// Usage - 100% type-safe, runtime validated
const ipc = createIpcProxy<IpcContract>(window.electronAPI)

const result = await ipc.locations.getAll({ warehouseId: 'WH-001' })
// ↑ Compile-time type checking + Runtime Zod validation
// Returns: Result<LocationsData, AppError>

// Benefits:
// - Single source of truth (no type duplication)
// - Compile-time type safety (TypeScript)
// - Runtime validation (Zod)
// - Lightweight (~50kb vs 2MB for tRPC)
// - Designed for IPC (not HTTP)
// - Easy to test (mock contract)
// - Refactor-safe (contract break = compile error)

Backend (Electron Main):

// backend/ipc/handlers.ts
import { ipcHandler } from './ipc-handler'
import { ipcContract } from '../../shared/ipc/contract'

ipcHandler.handle(ipcContract.locations.getAll, async (input) => {
  // Input already validated by Zod
  const data = await getLocationsByWarehouse(input?.warehouseId)
  return success(data) // Output validated by Zod
})

8. Functional Backend Architecture (No Services)

Backend should use pure functions and composition, no service layer:

// ❌ BAD: Service layer (delete this)
// services/analysis-service.ts
export const performABCAnalysis = (...args) => {
  return runABCAnalysis(...args)  // Useless wrapper!
}

// ✅ GOOD: Direct function calls
// backend/analysis/abc.ts
import { getProductMovementTotals } from '../database/queries'
import type { Result } from '../types/result'

export const analyzeABC = (
  warehouseId: string,
  dateFrom?: string,
  dateTo?: string
): Result<ABCAnalysisResult, DatabaseError> => {
  // Pure function, no side effects
  const movements = getProductMovementTotals(warehouseId, 'outbound', dateFrom, dateTo)

  if (movements.length === 0) {
    return failure(databaseError('QUERY_FAILED', 'abc_analysis', 'No movements found'))
  }

  const analysis = computeABCClassification(movements)
  return success(analysis)
}

// Usage: Direct import, no service layer
import { analyzeABC } from './backend/analysis/abc'
const result = analyzeABC(warehouseId)

Plugin Registry - Pure Functions:

// ❌ BAD: Global mutable state
export const registry: PluginRegistry = {}
export const registerPlugin = (plugin: ImportPlugin): void => {
  registry[plugin.id] = plugin  // Mutation!
}

// ✅ GOOD: Pure functions with readonly data
// backend/import/plugins/config.ts
import { genericExcelPlugin } from './generic-excel'
import { mockDataGeneratorPlugin } from './mock-data-generator'

const defaultPlugins: Readonly<Record<string, ImportPlugin>> = {
  [genericExcelPlugin.id]: genericExcelPlugin,
  [mockDataGeneratorPlugin.id]: mockDataGeneratorPlugin,
} as const

export const getPlugin = (id: string): ImportPlugin | undefined =>
  defaultPlugins[id]

export const listPlugins = (): readonly ImportPlugin[] =>
  Object.values(defaultPlugins)

export const withCustomPlugin = (
  plugin: ImportPlugin
): Readonly<Record<string, ImportPlugin>> => ({
  ...defaultPlugins,
  [plugin.id]: plugin,
})

Generic Database Operations:

// ❌ BAD: 20 identical functions
export const insertWarehouses = (warehouses: readonly any[]): number => {
  return bulkInsert(warehousesTable, warehouses, (w) => ({...}), 'Warehouses')
}
export const insertUsers = (users: readonly any[]): number => {
  return bulkInsert(usersTable, users, (u) => ({...}), 'Users')
}
// ... 18 more

// ✅ GOOD: Single generic function with Zod validation
import type { Result } from '../types/result'
import { warehouseSchema } from '../schemas/warehouse'

export const bulkInsertValidated = <T>(
  table: Table,
  schema: z.ZodSchema<T>,
  data: readonly T[]
): Result<number, ValidationError> => {
  // Validate all data first
  const validation = schema.array().safeParse(data)
  if (!validation.success) {
    return failure(validationError('BULK_INSERT_FAILED', validation.error))
  }

  // Use Drizzle transaction
  const inserted = db.transaction((items: readonly T[]) => {
    return items.reduce((count, item) => {
      try {
        db.insert(table).values(item).run()
        return count + 1
      } catch (error) {
        // Log and return count (don't throw)
        logError('INSERT_FAILED', { table: table[Symbol.for('name')], error })
        return count
      }
    }, 0)
  })(validation.data)

  return success(inserted)
}

// Usage - type-safe and validated
const result = bulkInsertValidated(warehousesTable, warehouseSchema, warehouses)

---

Affected Areas

Frontend

• All route components (locations.tsx, zones.tsx, sectors.tsx, etc.)
• useBackend() hook
• use-locations.ts (all data fetching hooks)
• All (window as any).electronAPI accesses
• Error handling throughout frontend
• Navigation/redirects

Backend

• Remove all src/backend/services/*.ts (fake service layer)
• src/backend/import/plugins/registry.ts (global mutable state)
• src/backend/import/loader.ts (code duplication, validation)
• src/backend/import/mapping-service.ts (TODO stubs)
• src/backend/database/index.ts (singleton, type safety)
• src/backend/analysis/*.ts (runtime requires)
• All console.log statements (replace with structured logging)
• All error handling (add Result types)

---

Implementation Plan

Phase 1: Foundation

1. Define core types

   • Result<T, E> type
   • AppError type
   • Error domains and codes

2. Create Zod IPC contract

   • Define all Zod schemas for entities (Location, Zone, Sector, etc.)
   • Create ipcContract with input/output schemas for all channels
   • Add type inference utilities (IpcInput, IpcOutput)
   • Implement createIpcProxy() with Zod validation

3. Create IPC wrapper

   • Implement safeIpcCall() with retry logic
   • Add structured logging
   • Replace all (window as any).electronAPI accesses with ipc proxy

Phase 2: Backend Refactoring (Functional Approach)

1. Remove fake service layer

   • Delete all src/backend/services/*.ts files
   • Move functions directly where they're needed
   • No intermediate "service" layer

2. Replace global mutable state

   • Convert plugin registry to pure functions
   • Pass plugins as parameters instead of global mutation
   • Create getPlugins() function that returns readonly list

3. Add data validation

   • Create Zod schemas for all entities
   • Validate at boundaries (IPC, database, file parsing)
   • Return Result<T, AppError> from all operations

4. Eliminate code duplication

   • Create generic bulkInsert with entity schemas
   • Use generic types instead of repeating for each entity
   • Reduce loader.ts from 750+ lines to <200 lines

5. Fix database singleton

   • Replace global singleton with dependency injection
   • Pass database connection as parameter
   • Remove any types, use proper Drizzle types

6. Replace runtime requires

   • Convert require() to static ES6 imports
   • Make dependencies explicit and type-checked

7. Remove TODO stubs

   • Either implement or remove saveMappingPreset/loadMappingPreset
   • No half-implemented features in production

Phase 3: Frontend Hooks

1. Simplify hooks

   • Remove "god hook" anti-pattern
   • Use IPC proxy directly through ipc object
   • Each hook calls ipc.entities.xxx() directly
   • Clear contract: ipc.locations.getAll({ warehouseId })

2. Fix navigation

   • Remove useEffect navigation
   • Use router loaders or conditional rendering

Phase 4: Error Handling

1. Error boundaries

   • Create error boundary components
   • Add user-friendly error displays

2. Monitoring/logging

   • Structured error logging
   • Development vs production behavior

---

Open Questions

1. Migration Strategy

   • Should we migrate incrementally or all at once?
   • How do we handle existing code during migration?

2. Backward Compatibility

   • Do we need to maintain existing APIs during transition?
   • What's our deprecation policy?

3. Testing Strategy

   • How do we test the new error handling patterns?
   • Do we need property-based testing?

4. User Experience

   • How do we display recoverable vs non-recoverable errors to users?
   • What retry behavior do we expose in the UI?

---

Related Issues

• Data mixing between warehouses (partially addressed)
• Dashboard vs detailed pages data inconsistency
• Mock data generation architecture

---

References

• Railway-Oriented Programming
• Result Type Pattern
• React Error Boundaries
• Functional Programming Patterns

[codewizdave]

Progress Update (2025-02-04)

✅ Completed (~35-40%)

Type Safety & IPC (Issues #4, #10, #14)

• Created TypedElectronAPI interface with full type definitions
• Removed ALL (window as any) casts from codebase
• Added Zod schemas for all entities in shared/schemas/entities.ts
• Implemented Zod validation on all IPC handlers
• Queries now return properly typed data

Error Handling (Issue #5)

• Implemented Result<T, AppError> pattern throughout codebase
• All IPC handlers return Result<T, AppError> with proper error structure
• Created helper functions (wrapResult, wrapError, handleAsync)
• Errors properly propagated to React Query hooks

Architecture (Issue #3, #6)

• Removed obsolete use-backend.ts "god hook" (558 lines)
• Created dedicated hooks: useLocations, useWarehouses, useZones, etc.
• Migrated all routes to use typed IPC through hooks
• Added IPC aliases in main.cjs for new API

Recent Commits:

• bb98d40 - Fix lastUpdated/updatedAt field mismatch in queries
• 391786a - Fix import paths and remove duplicate declarations
• 847c17d - Add generateMockData to typed API
• eaf3bcb - Finalize typing with proper types and Zod validation
• 31dc8de - Fix singleton pattern and add Result types
• d3b4efe - Complete migration to typed IPC
• 089e4fb - Add typed IPC handlers for all remaining endpoints
• 08c1490 - Add comprehensive Zod schemas

❌ Remaining Work (~60-65%)

Critical Issues:

• Implement Backend Import System with SQLite and Plugin Architecture #1: Unsafe navigation patterns with useEffect
• #7: Global mutable state in plugin registry
• #8: Runtime require() instead of static imports
• #11: Massive code duplication in loader.ts (750+ lines)
• #12: God function loadToDatabase (131 lines)
• #13: TODO stubs in mapping-service.ts
• #15: Global singleton database pattern
• #17: Hard-coded string literals (need enums)
• #18: Replace console.log with structured logging

Status: Issue should remain OPEN. Significant progress made but core architectural issues remain.

[codewizdave]

Progress Update #2 (2025-02-04)

✅ Additional Progress (~42% complete)

Static Imports (Issue #8) ✅

• Converted all require() calls to static ES6 imports
• parser.ts: Added imports for 'fs' and 'path'
• schema.ts: Added import for 'path'
• queries.ts: Added getDbRaw import, removed helper function

Plugin Registry (Issue #7) ✅

• Already refactored to pure functional approach
• Uses readonly defaultPlugins
• Pure functions: getPlugin(), listPlugins(), withPlugin()

Dynamic Imports (Issue #9) ✅

• No dynamic imports found - already cleaned up

TODO Stubs (Issue #13) ✅

• saveMappingPreset/loadMappingPreset functions removed
• mapping-service.ts is clean

Naming Conventions (Issue #19) ✅

• Created comprehensive docs/NAMING_CONVENTIONS.md
• Covers files, variables, functions, React, database, IPC, tests
• Includes summary table and anti-patterns

Recent Commits:

• 625fde0 - Convert runtime require() to static imports
• 91d49be - Add comprehensive naming conventions guide

📋 Remaining (~58%)

High Priority:

• Implement Backend Import System with SQLite and Plugin Architecture #1: Unsafe navigation with useEffect
• #11: Code duplication in loader.ts (750+ lines)
• #12: God function loadToDatabase (131 lines)
• #15: Global database singleton
• #16: Silent error handling

Medium Priority:

• Console.log replacement (logger exists, needs usage)
• Apply naming conventions throughout codebase

Status: Good progress! Core type safety and architecture issues resolved. Remaining work is mostly refactoring and cleanup.

Co-Authored-By: Claude Sonnet 4.5 noreply@anthropic.com

[roxdavirox]

The Result<T, E> pattern you're sketching is exactly what @roxdavirox/fp-core ships — no need to implement it from scratch. It has Ok, Err, mapResult, flatMap (your chain), match, tryCatch, and fromPromise, all TypeScript-native with zero dependencies:

import { pipe, Ok, Err, flatMap, match, tryCatch } from '@roxdavirox/fp-core';

const safeParse = tryCatch(JSON.parse);

pipe(
  safeParse(rawInput),
  flatMap(validate),
  flatMap(persist),
  match(
    data  => ({ status: 'ok',    data }),
    error => ({ status: 'error', error }),
  ),
);