@xrmforge/typegen 0.8.3 → 0.8.5

package/docs/architecture/10-eslint-plugin.md

@@ -1,82 +1,82 @@
-# ESLint Plugin
-
-### 10.1 Installation
-
-```javascript
-// eslint.config.js (flat config, ESLint v9)
-import xrmforge from '@xrmforge/eslint-plugin';
-
-export default [
-  xrmforge.configs.recommended,
-  // or pick individual rules
-];
-```
-
-### 10.2 Rules
-
-#### no-xrm-page (error)
-
-Forbids the deprecated `Xrm.Page` API (removed in D365 v9.0+).
-
-```typescript
-// Bad
-Xrm.Page.getAttribute("name");
-
-// Good
-const form = executionContext.getFormContext();
-form.getAttribute("name");
-```
-
-#### no-magic-optionset (warn)
-
-Forbids raw numbers (>= 2) in comparisons with `.getValue()`.
-
-```typescript
-// Bad
-if (attr.getValue() === 595300000) { }
-
-// Good
-import { StatusCode } from '../generated/optionsets/account';
-if (attr.getValue() === StatusCode.Active) { }
-```
-
-#### no-sync-webapi (error)
-
-Forbids synchronous XMLHttpRequest (`new XMLHttpRequest()` and `.open()` with `async=false`).
-
-```typescript
-// Bad
-xhr.open("GET", url, false);
-
-// Good
-const data = await Xrm.WebApi.retrieveRecord("account", id);
-```
-
-#### require-error-handling (warn)
-
-Requires try/catch in exported async functions starting with "on" (event handlers).
-
-```typescript
-// Bad
-export async function onLoad(ctx) {
-  await fetch("/api");  // no error handling
-}
-
-// Good
-export async function onLoad(ctx) {
-  try { await fetch("/api"); }
-  catch (error) { console.error(error); }
-}
-```
-
-#### require-namespace (warn)
-
-Forbids direct `window.X = ...` or `globalThis.X = ...` assignments. Module exports with esbuild globalName should be used instead.
-
-```typescript
-// Bad
-window.Contoso = { onLoad: function() {} };
-
-// Good
-export function onLoad(ctx: Xrm.Events.EventContext) {}
-```
+# ESLint Plugin
+
+### 10.1 Installation
+
+```javascript
+// eslint.config.js (flat config, ESLint v9)
+import xrmforge from '@xrmforge/eslint-plugin';
+
+export default [
+  xrmforge.configs.recommended,
+  // or pick individual rules
+];
+```
+
+### 10.2 Rules
+
+#### no-xrm-page (error)
+
+Forbids the deprecated `Xrm.Page` API (removed in D365 v9.0+).
+
+```typescript
+// Bad
+Xrm.Page.getAttribute("name");
+
+// Good
+const form = executionContext.getFormContext();
+form.getAttribute("name");
+```
+
+#### no-magic-optionset (warn)
+
+Forbids raw numbers (>= 2) in comparisons with `.getValue()`.
+
+```typescript
+// Bad
+if (attr.getValue() === 595300000) { }
+
+// Good
+import { StatusCode } from '../generated/optionsets/account';
+if (attr.getValue() === StatusCode.Active) { }
+```
+
+#### no-sync-webapi (error)
+
+Forbids synchronous XMLHttpRequest (`new XMLHttpRequest()` and `.open()` with `async=false`).
+
+```typescript
+// Bad
+xhr.open("GET", url, false);
+
+// Good
+const data = await Xrm.WebApi.retrieveRecord("account", id);
+```
+
+#### require-error-handling (warn)
+
+Requires try/catch in exported async functions starting with "on" (event handlers).
+
+```typescript
+// Bad
+export async function onLoad(ctx) {
+  await fetch("/api");  // no error handling
+}
+
+// Good
+export async function onLoad(ctx) {
+  try { await fetch("/api"); }
+  catch (error) { console.error(error); }
+}
+```
+
+#### require-namespace (warn)
+
+Forbids direct `window.X = ...` or `globalThis.X = ...` assignments. Module exports with esbuild globalName should be used instead.
+
+```typescript
+// Bad
+window.Contoso = { onLoad: function() {} };
+
+// Good
+export function onLoad(ctx: Xrm.Events.EventContext) {}
+```

package/docs/architecture/11-agent-md.md

@@ -1,38 +1,38 @@
-# AGENT.md System
-
-### 11.1 Purpose
-
-The AGENT.md is a scaffolded file that teaches AI coding assistants (Claude, ChatGPT, Copilot, Cursor) how to write optimal D365 form scripts using XrmForge. It is generated by `xrmforge init` and placed in the project root.
-
-### 11.2 Content Structure
-
-1. **Package overview** - What each @xrmforge package does
-2. **10 Rules: Always** - Fields Enum, OptionSet Enum, FormContext cast, EntityNames, parseLookup, select, createFormMock, module exports, Tabs/Sections enums, error handling
-3. **Rules: Never** - Raw strings, magic numbers, Xrm.Page, sync XHR, eval, window assignments
-4. **Before/After examples** - Field access, OptionSet comparison, testing
-5. **Pattern Recognition table** - Legacy pattern to XrmForge replacement mapping
-6. **OptionSet Enum creation guide** - How to create enums from magic numbers in legacy code
-7. **Testing with setupXrmMock** - Global Xrm mock pattern
-8. **Build commands** - xrmforge build, watch mode
-9. **@types/xrm Pitfalls** - Known issues and workarounds
-10. **File structure** - Expected project layout
-
-### 11.3 Template System
-
-The AGENT.md is stored as `src/scaffold/templates/AGENT.md` in the devkit package and loaded via `template-loader.ts` at scaffold time. No variable substitution needed (the file is static).
-
-### 11.4 KI Comparison Test Results
-
-Five AI models were tested converting legacy D365 JavaScript (account.js + lm_helper.js, 1,288 lines) to TypeScript with XrmForge:
-
-| Rank | Model | Score | Tool | Strength |
-|------|-------|-------|------|----------|
-| 1 | Claude Opus 4.6 | 42/50 | Claude Code | Most tests (62), best code structure |
-| 2 | Claude Sonnet 4.6 | 41/50 | Claude Code | Most bugs found (5), best DI approach |
-| 3 | Cursor Composer 2 | 35/50 | Cursor IDE | Recognized select() Node API issue |
-| 4 | ChatGPT GPT-4o | 30/50 | ChatGPT Web | Functional but less XrmForge-specific |
-| 5 | MS Copilot | 12/50 | Browser Chat | No workspace access, never saw AGENT.md |
-
-**Criteria (11, max 5 points each = 55 max):** Fields Enum usage, OptionSet Enums, FormContext typing, XrmForge helpers, module exports, tests present, test quality, error handling, code quality, bugs found, documentation.
-
-**Key finding:** No AI consistently used `@xrmforge/helpers` imports (select, parseLookup). This remains the biggest adoption gap.
+# AGENT.md System
+
+### 11.1 Purpose
+
+The AGENT.md is a scaffolded file that teaches AI coding assistants (Claude, ChatGPT, Copilot, Cursor) how to write optimal D365 form scripts using XrmForge. It is generated by `xrmforge init` and placed in the project root.
+
+### 11.2 Content Structure
+
+1. **Package overview** - What each @xrmforge package does
+2. **10 Rules: Always** - Fields Enum, OptionSet Enum, FormContext cast, EntityNames, parseLookup, select, createFormMock, module exports, Tabs/Sections enums, error handling
+3. **Rules: Never** - Raw strings, magic numbers, Xrm.Page, sync XHR, eval, window assignments
+4. **Before/After examples** - Field access, OptionSet comparison, testing
+5. **Pattern Recognition table** - Legacy pattern to XrmForge replacement mapping
+6. **OptionSet Enum creation guide** - How to create enums from magic numbers in legacy code
+7. **Testing with setupXrmMock** - Global Xrm mock pattern
+8. **Build commands** - xrmforge build, watch mode
+9. **@types/xrm Pitfalls** - Known issues and workarounds
+10. **File structure** - Expected project layout
+
+### 11.3 Template System
+
+The AGENT.md is stored as `src/scaffold/templates/AGENT.md` in the devkit package and loaded via `template-loader.ts` at scaffold time. No variable substitution needed (the file is static).
+
+### 11.4 KI Comparison Test Results
+
+Five AI models were tested converting legacy D365 JavaScript (account.js + lm_helper.js, 1,288 lines) to TypeScript with XrmForge:
+
+| Rank | Model | Score | Tool | Strength |
+|------|-------|-------|------|----------|
+| 1 | Claude Opus 4.6 | 42/50 | Claude Code | Most tests (62), best code structure |
+| 2 | Claude Sonnet 4.6 | 41/50 | Claude Code | Most bugs found (5), best DI approach |
+| 3 | Cursor Composer 2 | 35/50 | Cursor IDE | Recognized select() Node API issue |
+| 4 | ChatGPT GPT-4o | 30/50 | ChatGPT Web | Functional but less XrmForge-specific |
+| 5 | MS Copilot | 12/50 | Browser Chat | No workspace access, never saw AGENT.md |
+
+**Criteria (11, max 5 points each = 55 max):** Fields Enum usage, OptionSet Enums, FormContext typing, XrmForge helpers, module exports, tests present, test quality, error handling, code quality, bugs found, documentation.
+
+**Key finding:** No AI consistently used `@xrmforge/helpers` imports (select, parseLookup). This remains the biggest adoption gap.

package/docs/architecture/12-xrm-pitfalls.md

@@ -1,14 +1,14 @@
-# @types/xrm Pitfalls
-
-Known issues when working with `@types/xrm`:
-
-| Issue | Wrong | Correct |
-|-------|-------|---------|
-| Form interface | `interface extends Xrm.FormContext` | `extends Omit<Xrm.FormContext, 'getAttribute' \| 'getControl'>` |
-| AlertDialogResponse | `Xrm.Navigation.AlertDialogResponse` | `Xrm.Async.PromiseLike<void>` (type does not exist) |
-| ConfirmDialogResponse | `Xrm.Navigation.ConfirmDialogResponse` | `Xrm.Navigation.ConfirmResult` (type does not exist) |
-| setNotification | `setNotification(message)` | `setNotification(message, uniqueId)` (requires 2 args) |
-| openFile | `openFile({ fileName, ... })` | Must include `fileSize` property in FileDetails |
-| SubmitMode | `Xrm.Attributes.SubmitMode` | `Xrm.SubmitMode` |
-| const enum in .d.ts | `const enum` in `.d.ts` files | Use `const enum` in `.ts` ES modules (typegen 0.8.0+ generates `.ts` files, resolving this issue) |
-| Grid.refresh() | `grid.refresh()` | `(grid as any).refresh()` (not typed in @types/xrm) |
+# @types/xrm Pitfalls
+
+Known issues when working with `@types/xrm`:
+
+| Issue | Wrong | Correct |
+|-------|-------|---------|
+| Form interface | `interface extends Xrm.FormContext` | `extends Omit<Xrm.FormContext, 'getAttribute' \| 'getControl'>` |
+| AlertDialogResponse | `Xrm.Navigation.AlertDialogResponse` | `Xrm.Async.PromiseLike<void>` (type does not exist) |
+| ConfirmDialogResponse | `Xrm.Navigation.ConfirmDialogResponse` | `Xrm.Navigation.ConfirmResult` (type does not exist) |
+| setNotification | `setNotification(message)` | `setNotification(message, uniqueId)` (requires 2 args) |
+| openFile | `openFile({ fileName, ... })` | Must include `fileSize` property in FileDetails |
+| SubmitMode | `Xrm.Attributes.SubmitMode` | `Xrm.SubmitMode` |
+| const enum in .d.ts | `const enum` in `.d.ts` files | Use `const enum` in `.ts` ES modules (typegen 0.8.0+ generates `.ts` files, resolving this issue) |
+| Grid.refresh() | `grid.refresh()` | `(grid as any).refresh()` (not typed in @types/xrm) |

package/docs/architecture/13-helpers.md

@@ -1,50 +1,50 @@
-# @xrmforge/helpers Package
-
-### 13.1 Problem
-
-The previous approach used a `/helpers` subpath export on `@xrmforge/typegen`. This was confusing because typegen is a Node.js code generation tool, while helpers are browser-safe runtime utilities. The subpath `@xrmforge/typegen/helpers` was non-obvious and AI coding assistants consistently failed to discover it.
-
-### 13.2 Solution
-
-A standalone `@xrmforge/helpers` package consolidates all browser-safe runtime code. Zero Node.js dependencies. Clean, discoverable import path:
-
-```typescript
-// Import from the dedicated helpers package
-import { select, parseLookup, typedForm } from '@xrmforge/helpers';
-```
-
-### 13.3 Exports
-
-**Web API Helpers:**
-- `select(...fields: string[]): string` - Builds `?$select=field1,field2`
-- `selectExpand(fields: string[], expand: string): string` - Builds `?$select=...&$expand=...`
-- `parseLookup(response: Record<string, unknown>, fieldName: string): LookupValue | null` - Parses `_fieldname_value` with OData annotations
-- `parseLookups(response: Record<string, unknown>, fieldName: string): LookupValue[]` - Multi-value lookup parsing
-- `parseFormattedValue(response: Record<string, unknown>, fieldName: string): string | null` - Extracts `@OData.Community.Display.V1.FormattedValue`
-
-**Xrm Constants (8 const enums):**
-- DisplayState, FormNotificationLevel, RequiredLevel, SubmitMode, SaveMode, ClientType, ClientState, OperationType
-
-**typedForm() Proxy:**
-- `typedForm<TForm>(formContext)` - Returns a proxy where `form.name` delegates to `getAttribute('name')`
-- GET trap: Property access delegates to getAttribute(); `$context` returns raw FormContext; `$control(name)` returns getControl()
-- SET trap: Throws TypeError forcing `.setValue()` usage
-- HAS trap: Checks if attribute exists on the form
-
-**Action/Function Runtime:**
-- `createBoundAction(entityName, actionName)` - Creates a bound action executor
-- `executeRequest(request)` - Executes an Organization Request via Xrm.WebApi.online.execute
-- `withProgress(message, fn)` - Wraps an async operation with Xrm.Utility.showProgressIndicator
-
-### 13.4 Migration
-
-The old import path `@xrmforge/typegen/helpers` has been removed. Update all imports:
-
-```typescript
-// Old (removed)
-import { select } from '@xrmforge/typegen/helpers';
-import { typedForm } from '@xrmforge/formhelpers';
-
-// New
-import { select, typedForm } from '@xrmforge/helpers';
-```
+# @xrmforge/helpers Package
+
+### 13.1 Problem
+
+The previous approach used a `/helpers` subpath export on `@xrmforge/typegen`. This was confusing because typegen is a Node.js code generation tool, while helpers are browser-safe runtime utilities. The subpath `@xrmforge/typegen/helpers` was non-obvious and AI coding assistants consistently failed to discover it.
+
+### 13.2 Solution
+
+A standalone `@xrmforge/helpers` package consolidates all browser-safe runtime code. Zero Node.js dependencies. Clean, discoverable import path:
+
+```typescript
+// Import from the dedicated helpers package
+import { select, parseLookup, typedForm } from '@xrmforge/helpers';
+```
+
+### 13.3 Exports
+
+**Web API Helpers:**
+- `select(...fields: string[]): string` - Builds `?$select=field1,field2`
+- `selectExpand(fields: string[], expand: string): string` - Builds `?$select=...&$expand=...`
+- `parseLookup(response: Record<string, unknown>, fieldName: string): LookupValue | null` - Parses `_fieldname_value` with OData annotations
+- `parseLookups(response: Record<string, unknown>, fieldName: string): LookupValue[]` - Multi-value lookup parsing
+- `parseFormattedValue(response: Record<string, unknown>, fieldName: string): string | null` - Extracts `@OData.Community.Display.V1.FormattedValue`
+
+**Xrm Constants (8 const enums):**
+- DisplayState, FormNotificationLevel, RequiredLevel, SubmitMode, SaveMode, ClientType, ClientState, OperationType
+
+**typedForm() Proxy:**
+- `typedForm<TForm>(formContext)` - Returns a proxy where `form.name` delegates to `getAttribute('name')`
+- GET trap: Property access delegates to getAttribute(); `$context` returns raw FormContext; `$control(name)` returns getControl()
+- SET trap: Throws TypeError forcing `.setValue()` usage
+- HAS trap: Checks if attribute exists on the form
+
+**Action/Function Runtime:**
+- `createBoundAction(entityName, actionName)` - Creates a bound action executor
+- `executeRequest(request)` - Executes an Organization Request via Xrm.WebApi.online.execute
+- `withProgress(message, fn)` - Wraps an async operation with Xrm.Utility.showProgressIndicator
+
+### 13.4 Migration
+
+The old import path `@xrmforge/typegen/helpers` has been removed. Update all imports:
+
+```typescript
+// Old (removed)
+import { select } from '@xrmforge/typegen/helpers';
+import { typedForm } from '@xrmforge/formhelpers';
+
+// New
+import { select, typedForm } from '@xrmforge/helpers';
+```

package/docs/architecture/14-showcases.md

@@ -1,21 +1,21 @@
-# Showcases
-
-### 14.1 Markant WebResources (Production Showcase)
-
-Located in the XrmForge-Workspace repository under `docs/07_showcase/markant-webresources/`.
-
-- **30 WebResources** in `src/forms/` (account, contact, opportunity, lead, quote, email, task, etc.)
-- **1 shared library** (GDPR retention UI)
-- **9 test files** with 59 tests
-- **79 generated typings:** 25 form interfaces, 28 entity interfaces, 22 OptionSet files, 4 action executors
-- **esbuild build** via xrmforge.config.json (32 entries)
-- **Deploy script** (deploy.mjs) with @azure/identity auth, incremental deployment, hash-based change detection
-- **27 entities, 236 OptionSet enums, 95 form interfaces, 7 Custom API executors**
-
-### 14.2 LMApp WebResources (KI Comparison Showcase)
-
-Created during the KI comparison tests (Session 9). 18 legacy JavaScript form scripts (~8,400 lines) converted to TypeScript with XrmForge patterns.
-
-- **19 WebResources** with Fields Enums, EntityNames, OptionSet Enums
-- **84 tests** in 8 test files
-- **XrmForge-optimized:** All 10 AGENT.md rules applied (FormContext cast, Fields Enum, EntityNames, OptionSet Enums, shared getLookupObject, Tab Enums)
+# Showcases
+
+### 14.1 Markant WebResources (Production Showcase)
+
+Located in the XrmForge-Workspace repository under `docs/07_showcase/markant-webresources/`.
+
+- **30 WebResources** in `src/forms/` (account, contact, opportunity, lead, quote, email, task, etc.)
+- **1 shared library** (GDPR retention UI)
+- **9 test files** with 59 tests
+- **79 generated typings:** 25 form interfaces, 28 entity interfaces, 22 OptionSet files, 4 action executors
+- **esbuild build** via xrmforge.config.json (32 entries)
+- **Deploy script** (deploy.mjs) with @azure/identity auth, incremental deployment, hash-based change detection
+- **27 entities, 236 OptionSet enums, 95 form interfaces, 7 Custom API executors**
+
+### 14.2 LMApp WebResources (KI Comparison Showcase)
+
+Created during the KI comparison tests (Session 9). 18 legacy JavaScript form scripts (~8,400 lines) converted to TypeScript with XrmForge patterns.
+
+- **19 WebResources** with Fields Enums, EntityNames, OptionSet Enums
+- **84 tests** in 8 test files
+- **XrmForge-optimized:** All 10 AGENT.md rules applied (FormContext cast, Fields Enum, EntityNames, OptionSet Enums, shared getLookupObject, Tab Enums)

package/docs/architecture/15-ci-cd.md

@@ -1,49 +1,49 @@
-# CI/CD
-
-### 15.1 GitHub Actions CI (`.github/workflows/ci.yml`)
-
-**Triggers:** Push to main, Pull Requests against main.
-
-**Matrix:** Node 20, Node 22 on ubuntu-latest.
-
-**Steps:**
-1. Checkout
-2. Setup pnpm (from packageManager field)
-3. Setup Node.js (matrix version)
-4. `pnpm install --frozen-lockfile`
-5. `pnpm lint`
-6. `pnpm -r exec tsc --noEmit` (typecheck all packages)
-7. `pnpm build`
-8. `pnpm test`
-9. Coverage (Node 22 only): `npx vitest run --coverage` in typegen
-
-### 15.2 Release Workflow (`.github/workflows/release.yml`)
-
-**Triggers:** After successful CI on push to main.
-
-**Steps:**
-1. Checkout, setup pnpm, setup Node 22
-2. `pnpm install --frozen-lockfile`
-3. `pnpm build`
-4. Changesets action: creates Release PR or publishes to npm
-
-**Publish command:** `pnpm release` = `turbo run build && changeset publish`
-
-### 15.3 Turbo Pipeline
-
-```
-build:      dependsOn: [^build], outputs: [dist/**]
-test:       dependsOn: [build]
-typecheck:  dependsOn: [^build]
-lint:       (no dependencies)
-dev:        cache: false, persistent: true
-clean:      cache: false
-```
-
-### 15.4 Changesets
-
-Configured for public npm access, auto-update internal dependencies on patch level. Publish requires NPM_TOKEN secret.
-
-### 15.5 Publishing Order
-
-Due to internal dependencies: typegen first, then devkit, then cli. Must use `pnpm publish` (not `npm publish`) to resolve `workspace:*` references to real versions.
+# CI/CD
+
+### 15.1 GitHub Actions CI (`.github/workflows/ci.yml`)
+
+**Triggers:** Push to main, Pull Requests against main.
+
+**Matrix:** Node 20, Node 22 on ubuntu-latest.
+
+**Steps:**
+1. Checkout
+2. Setup pnpm (from packageManager field)
+3. Setup Node.js (matrix version)
+4. `pnpm install --frozen-lockfile`
+5. `pnpm lint`
+6. `pnpm -r exec tsc --noEmit` (typecheck all packages)
+7. `pnpm build`
+8. `pnpm test`
+9. Coverage (Node 22 only): `npx vitest run --coverage` in typegen
+
+### 15.2 Release Workflow (`.github/workflows/release.yml`)
+
+**Triggers:** After successful CI on push to main.
+
+**Steps:**
+1. Checkout, setup pnpm, setup Node 22
+2. `pnpm install --frozen-lockfile`
+3. `pnpm build`
+4. Changesets action: creates Release PR or publishes to npm
+
+**Publish command:** `pnpm release` = `turbo run build && changeset publish`
+
+### 15.3 Turbo Pipeline
+
+```
+build:      dependsOn: [^build], outputs: [dist/**]
+test:       dependsOn: [build]
+typecheck:  dependsOn: [^build]
+lint:       (no dependencies)
+dev:        cache: false, persistent: true
+clean:      cache: false
+```
+
+### 15.4 Changesets
+
+Configured for public npm access, auto-update internal dependencies on patch level. Publish requires NPM_TOKEN secret.
+
+### 15.5 Publishing Order
+
+Due to internal dependencies: typegen first, then devkit, then cli. Must use `pnpm publish` (not `npm publish`) to resolve `workspace:*` references to real versions.

package/docs/architecture/16-technical-debt.md

@@ -1,17 +1,17 @@
-# Technical Debt
-
-### 16.1 Known Issues
-
-| Issue | Status | Priority |
-|-------|--------|----------|
-| parseLookup/select not adopted by AI assistants | Open | High |
-| release.yml double runs (CI triggers release, release re-triggers CI) | Open | Low |
-| No integration tests against live Dataverse | Open (OE-4) | Medium |
-| @xrmforge/webapi has no Action/Function support | Accepted | Low |
-| devDependency versions in scaffolded package.json are pinned to old versions | Open | Low |
-
-### 16.2 Accepted Limitations
-
-- **const enum limitation:** Resolved in typegen 0.8.0. Generated output is now `.ts` ES modules, so `const enum` works directly with vitest and other test frameworks.
-- **Grid.refresh() requires `as any`:** Not typed in @types/xrm.
-- **Single solution per entity:** If an entity appears in multiple solutions, it is only generated once.
+# Technical Debt
+
+### 16.1 Known Issues
+
+| Issue | Status | Priority |
+|-------|--------|----------|
+| parseLookup/select not adopted by AI assistants | Open | High |
+| release.yml double runs (CI triggers release, release re-triggers CI) | Open | Low |
+| No integration tests against live Dataverse | Open (OE-4) | Medium |
+| @xrmforge/webapi has no Action/Function support | Accepted | Low |
+| devDependency versions in scaffolded package.json are pinned to old versions | Open | Low |
+
+### 16.2 Accepted Limitations
+
+- **const enum limitation:** Resolved in typegen 0.8.0. Generated output is now `.ts` ES modules, so `const enum` works directly with vitest and other test frameworks.
+- **Grid.refresh() requires `as any`:** Not typed in @types/xrm.
+- **Single solution per entity:** If an entity appears in multiple solutions, it is only generated once.

package/docs/architecture/17-roadmap.md

@@ -1,25 +1,25 @@
-# Roadmap
-
-### 17.1 Next Steps (Priority Order)
-
-1. **parseLookup/select Adoption** - Improve AGENT.md examples so AI assistants consistently use `/helpers` imports
-2. **LMApp Showcase regeneration** - With latest releases (testing@0.2.0, devkit@0.4.0 with improved AGENT.md)
-3. **KI Battle Round 3** - Re-test Sonnet vs Opus after improvements to measure progress
-4. **Documentation website** - xrmforge.dev or xrmforge.io (OE-3)
-
-### 17.2 Open Decisions
-
-| ID | Decision | Status |
-|----|----------|--------|
-| OE-1 | npm scope availability (@xrmforge) | Open |
-| OE-2 | GitHub org vs personal repo | Decided: personal (juergenbeck/XrmForge) |
-| OE-3 | Documentation domain (xrmforge.dev or .io) | Open |
-| OE-4 | Dataverse test environment for integration tests | Open |
-| OE-5 | Publisher prefix and solution name for PCF/WebResource tests | Open |
-
-### 17.3 Future Possibilities
-
-- Relationship Names const enum (OE-7, low priority)
-- @xrmforge/webapi with Action/Function support (reuse DataverseHttpClient)
-- Plugin system for custom generators and type mappings
-- Server-side generation (Custom API in Dataverse)
+# Roadmap
+
+### 17.1 Next Steps (Priority Order)
+
+1. **parseLookup/select Adoption** - Improve AGENT.md examples so AI assistants consistently use `/helpers` imports
+2. **LMApp Showcase regeneration** - With latest releases (testing@0.2.0, devkit@0.4.0 with improved AGENT.md)
+3. **KI Battle Round 3** - Re-test Sonnet vs Opus after improvements to measure progress
+4. **Documentation website** - xrmforge.dev or xrmforge.io (OE-3)
+
+### 17.2 Open Decisions
+
+| ID | Decision | Status |
+|----|----------|--------|
+| OE-1 | npm scope availability (@xrmforge) | Open |
+| OE-2 | GitHub org vs personal repo | Decided: personal (juergenbeck/XrmForge) |
+| OE-3 | Documentation domain (xrmforge.dev or .io) | Open |
+| OE-4 | Dataverse test environment for integration tests | Open |
+| OE-5 | Publisher prefix and solution name for PCF/WebResource tests | Open |
+
+### 17.3 Future Possibilities
+
+- Relationship Names const enum (OE-7, low priority)
+- @xrmforge/webapi with Action/Function support (reuse DataverseHttpClient)
+- Plugin system for custom generators and type mappings
+- Server-side generation (Custom API in Dataverse)

package/docs/architecture/18-design-principles.md

@@ -1,22 +1,22 @@
-# Design Principles
-
-The 18 design principles that govern all XrmForge development:
-
-1. **Extend, don't replace** - Types build on @types/xrm, never override them.
-2. **TypeScript all the way** - 100% TypeScript-native. No .NET, no ADAL.
-3. **Code must build** - Every work step ends with green build + tests.
-4. **Research before speed** - Investigate, compare, decide, then implement. Never guess.
-5. **No module without basics** - Error handling, logging, unit tests, JSDoc on all public APIs.
-6. **Monorepo discipline** - Each package standalone, no circular deps, barrel exports.
-7. **Enterprise resilience** - Retry + exponential backoff, rate-limit awareness, token caching, read-only default.
-8. **esbuild-first, webpack-compatible** - Default: esbuild (fast). webpack stays supported. IIFE output for D365.
-9. **MSAL-only authentication** - Only @azure/identity (no legacy ADAL). Three flows: client credentials, browser, device code.
-10. **Review required** - After every step, immediate critical review (6 dimensions). No asking if review is wanted.
-11. **Session state required** - session-state.md updated, changelog written, open questions tracked.
-12. **No half measures** - Every step completed fully: green build + tests + review before next step.
-13. **Informed architecture decisions** - Research, compare, recommend with pros/cons, get decision, persist.
-14. **Abstraction over vendor lock-in** - External dependencies behind interfaces (parser, auth, bundler).
-15. **Dual-language labels** - Primary language (1033/English) for identifiers, secondary in JSDoc. German umlauts transliterated.
-16. **Review with research and live verification** - Internet research, live D365 verification, production code checks, cite sources.
-17. **Challenge postponement** - "Later" check: Will it get harder? API contract? Real effort? Technical reasons?
-18. **Read-only default for Dataverse access** - DataverseHttpClient defaults to readOnly: true. Write access is an explicit opt-in.
+# Design Principles
+
+The 18 design principles that govern all XrmForge development:
+
+1. **Extend, don't replace** - Types build on @types/xrm, never override them.
+2. **TypeScript all the way** - 100% TypeScript-native. No .NET, no ADAL.
+3. **Code must build** - Every work step ends with green build + tests.
+4. **Research before speed** - Investigate, compare, decide, then implement. Never guess.
+5. **No module without basics** - Error handling, logging, unit tests, JSDoc on all public APIs.
+6. **Monorepo discipline** - Each package standalone, no circular deps, barrel exports.
+7. **Enterprise resilience** - Retry + exponential backoff, rate-limit awareness, token caching, read-only default.
+8. **esbuild-first, webpack-compatible** - Default: esbuild (fast). webpack stays supported. IIFE output for D365.
+9. **MSAL-only authentication** - Only @azure/identity (no legacy ADAL). Three flows: client credentials, browser, device code.
+10. **Review required** - After every step, immediate critical review (6 dimensions). No asking if review is wanted.
+11. **Session state required** - session-state.md updated, changelog written, open questions tracked.
+12. **No half measures** - Every step completed fully: green build + tests + review before next step.
+13. **Informed architecture decisions** - Research, compare, recommend with pros/cons, get decision, persist.
+14. **Abstraction over vendor lock-in** - External dependencies behind interfaces (parser, auth, bundler).
+15. **Dual-language labels** - Primary language (1033/English) for identifiers, secondary in JSDoc. German umlauts transliterated.
+16. **Review with research and live verification** - Internet research, live D365 verification, production code checks, cite sources.
+17. **Challenge postponement** - "Later" check: Will it get harder? API contract? Real effort? Technical reasons?
+18. **Read-only default for Dataverse access** - DataverseHttpClient defaults to readOnly: true. Write access is an explicit opt-in.