* refactor: eliminate duplicate code in JSON codecs by implementing encode/decode in terms of writeTo/readFrom Refactored all JSON codec modules to implement encode and decode functions in terms of writeTo and readFrom, eliminating code duplication and improving maintainability. Design Decisions: - encode functions now create a MemoryStream/Utf8JsonWriter, call writeTo, then parse the stream into a JsonElement - decode functions now extract raw JSON text from JsonElement, create a Utf8JsonReader, and call readFrom - This ensures a single source of truth for serialization/deserialization logic, making future changes easier and reducing the risk of inconsistencies Benefits: - Eliminates duplicate serialization logic between encode/writeTo - Eliminates duplicate deserialization logic between decode/readFrom - Simplifies maintenance: fixes and enhancements only need to be made once - Maintains API compatibility: encode/decode still available for JsonElement workflows, writeTo/readFrom available for streaming scenarios Codecs Refactored: - NameCodec: encode now uses writeTo, decode now uses readFrom - PathCodec: encode now uses writeTo, decode now uses readFrom - FQNameCodec: encode now uses writeTo, decode now uses readFrom - LiteralCodec: encode already used writeTo, decode now uses readFrom All 392 tests passing after refactoring. * refactor: make codec methods consistently accept MorphirJsonOptions This refactoring ensures all codec methods can respond to configuration options for future flexibility and extensibility. Changes: - Added MorphirJsonOptions.getOrDefault helper for optional parameter handling - Added MorphirJsonOptions.tryDetectVersion for auto-detecting format versions - Refactored all codec modules (Name, Path, PackageName, ModulePath, FQName, Literal) to provide both *WithOptions and default versions of all methods - Updated all JsonConverters to accept MorphirJsonOptions in constructor - Added SemVer format support for Names (kebab-case strings vs arrays) - Updated read methods to auto-detect format (accepts both Classic and SemVer) - Updated all tests (392 tests passing) - Updated serialization.md documentation with new API examples API Pattern: Each codec now provides: - writeToWithOptions, readFromWithOptions, encodeWithOptions, decodeWithOptions (explicitly accept MorphirJsonOptions) - writeTo, readFrom, encode, decode (convenience methods using default v3 options) Benefits: - Future-proof: Easy to add new options beyond FormatVersion - Consistent: All codecs follow the same pattern - Backward compatible: Default methods work without breaking changes - Flexible: Options can be threaded through codec call chains - Type-safe: Full MorphirJsonOptions type instead of bare FormatVersion 🤖 Generated with [Claude Code](https://claude.com/claude-code) * feat: Add Thoth-inspired JSON codecs with version-aware encoding for IR types - Implement composable Encoder/Decoder types following Thoth.Json pattern - Add Encode module with composable encoder functions (unit, string, int, list, etc.) - Add Decode module with composable decoder functions with Result-based error handling - Implement generic TypeCodec parameterized by attribute encoders/decoders - Add RawTypeCodec as specialized instance for Type<unit> - Support version-aware tag naming (v1: lowercase, v2+: PascalCase) - Enable cross-version JSON decoding (v1/v2/v3 formats all decodable) - Refactor IR types to namespace level (Pattern, Type, Literal, Value) - Add type aliases following morphir-elm conventions (RawType, TypedValue, etc.) - Add comprehensive test suite for RawTypeCodec (roundtrip, version-specific, structure) Architecture: - Core types remain simple: Encoder<'a> = 'a -> JsonElement - Version control via encodeWithOptions/decodeWithOptions functions - Generic codecs eliminate duplication across attribute types - Fully composable and type-safe design 🤖 Generated with [Claude Code](https://claude.com/claude-code) * refactor: Move Type and Field to namespace level in companion object pattern Refactored Type.fs to follow the companion object pattern where Type<'attributes> and Field<'attributes> are at the namespace level while keeping Constructors, TypeSpecification, TypeDefinition, and DerivedTypeDetails nested in the Type module. This change: - Moves Type<'attributes> and Field<'attributes> to Morphir.IR.Classic namespace - Keeps Type.Constructors, Type.TypeSpecification, Type.TypeDefinition, Type.DerivedTypeDetails nested - Updates all references in Aliases.fs, Module.fs, and DSL/Modules.fs - Maintains consistency with Pattern and Literal which already use this pattern - Type and Field can now be referenced without Type. prefix - Nested types require Type. qualification (e.g., Type.TypeDefinition) This pattern provides cleaner usage for the most commonly used types while maintaining logical grouping for specification and definition types. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * feat: Add AutoOpen TypeExtensions module for F# and C# compatibility Created TypeExtensions AutoOpen module with both F# type extensions and C#-visible extension methods for Type. This provides the Arrow fluent method and ^-> operator for creating function types. Changes: - Moved Arrow extension methods from Type module to AutoOpen TypeExtensions - Added F# type extensions (natural F# style) - Added C#-visible extensions using [<Extension>] attribute - Added ^-> operator for function type creation (F#-only) - Fixed TypeTests to use AutoOpen extensions instead of importing from Type module Benefits: - Extensions automatically available when Morphir.IR.Classic is opened - Natural F# syntax with type extensions - C# developers can use .Arrow() extension method - No need to explicitly open Type module for extensions 🤖 Generated with [Claude Code](https://claude.com/claude-code) * feat: Add AutoOpen PatternExtensions and fix test compilation errors Created AutoOpen PatternExtensions module for Pattern type extensions (.As and .Cons) with both F# and C# support. Fixed compilation errors in DSL test files. Changes: - Added PatternExtensions AutoOpen module in Pattern.fs - Moved As() and Cons() extensions from Pattern module to AutoOpen - Added C#-visible extension methods using [<Extension>] attribute - Fixed TypesTests.fs: removed manual ^-> import (now from AutoOpen) - Fixed LiteralsTests.fs: changed to open Morphir.IR.Classic, qualified literal helpers with Literal. - Fixed PatternsTests.fs: added open Morphir.IR.Classic for PatternExtensions Benefits: - Pattern extensions automatically available when Morphir.IR.Classic is opened - C# developers can use .As() and .Cons() extension methods - Consistent pattern with TypeExtensions - TypesTests, LiteralsTests, PatternsTests now compile successfully 🤖 Generated with [Claude Code](https://claude.com/claude-code) * fix: Remove invalid Literal module open in ValuesTests Changed ValuesTests to open Morphir.IR.Classic instead of trying to open the Morphir.IR.Classic.Literal module (which is RequireQualifiedAccess). This fixes the FS0892 error. Remaining ValuesTests errors are type confusion issues between Type and Value that need investigation. Remaining errors: 26 (all in ValuesTests.fs) 🤖 Generated with [Claude Code](https://claude.com/claude-code) * refactor: Implement companion object pattern for Value and fix Name codec normalization - Moved Value, ValueSpecification, and ValueDefinition to namespace level - Added RequireQualifiedAccess to Value module for helper functions - Qualified Type pattern matches in TypeCodec to resolve name collisions - Fixed NameCodec to normalize segments to lowercase during deserialization - Added Morphir.Models.Tests to Nuke Test target - Removed invalid open statements across DSL and test files - Added case normalization test to prevent regression All 410 tests passing. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * feat: Integrate Morphir.IR.Pipeline.Tests into CI/CD pipeline - Added Morphir.IR.Pipeline and Morphir.IR.Pipeline.Tests to solution - Added MorphirIRPipelineTestsProject path to Build.cs - Updated Test target to run Pipeline tests (92 tests) - Updated failure checking to include pipeline test exit code - All 915 tests now passing (277 + 111 + 25 + 410 + 92) This completes the integration of PR #356's Pipeline tests which were previously not running as part of the CI/CD Test target. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * feat: Add Morphir.E2E.Tests to solution Added Morphir.E2E.Tests.csproj to the solution file. This project is used by the BuildE2ETests and TestE2E Nuke targets but was previously missing from the solution. All active projects are now included in the solution. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * chore: Remove obsolete F# test files from old migration Removed Morphir.Core.Tests.fsproj and all F# source files from tests/Morphir.Core.Tests/. These were from an old migration and are superseded by the current C# implementation in Morphir.Core.Tests.csproj. Test count reduced from 277 to 237 in Core.Tests as the 40 F# tests were duplicates. All 875 tests passing. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * chore: Add Nuke build project to solution Added build/_build.csproj to the solution file for better IDE integration and developer experience. This follows Nuke best practices by treating the build project as a first-class citizen of the codebase. Benefits: - Better IntelliSense and navigation for build scripts - Proper NuGet package restore in Visual Studio - Easier modification and testing of build targets 🤖 Generated with [Claude Code](https://claude.com/claude-code) * fix: Make optional Nuke parameters nullable to resolve CS8618 warnings Changed optional CLI parameters to nullable types: - Rid (string -> string?) - Only required for publish targets - ApiKey (string -> string?) - Only required for NuGet publish - PluginVersion (string -> string?) - Only required for proto plugin packaging Build now compiles with 0 warnings, 0 errors. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * docs: Add comprehensive F# backend design documentation Complete design documentation suite for F# code generation backend: **Core Documents:** - PRD: Product requirements, architecture, success criteria - README: Central hub with quick links and guidance - GitHub Issues: 10 detailed implementation issues (Epic + 9 phases) - Issue Creation Script: Automated GitHub issue creation **Strategic Planning:** - Migration Assessment: NATIVE/ADAPT/MIGRATE/SKIP/NEW framework for morphir-elm evaluation - SDK Library Plan: Morphir.SDK F# runtime library design - Maturity Milestones: Incremental value delivery roadmap (M0-M5) **Key Features:** - 13-week implementation timeline (Phase 0-8) - Maturity-based releases: 0.4.0-alpha → 0.4.0-beta → 0.4.0-rc → 0.4.0 - Fabulous.AST + Fantomas for code generation - Morphir.SDK F# library for runtime support - Systematic morphir-elm migration assessment - TDD approach with ≥80% coverage requirement **Architecture:** - Load (JSON → IR) → Transform (IR → AST) → Write (AST → F# files) - Pluggable transformation pipeline integration - AOT-compatible generated code - SDK type mapping: Maybe → Option, Result → Result, etc. All documentation cross-referenced and ready for implementation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) * docs: Update F# backend documentation with actual GitHub issue numbers - Updated README-fsharp-backend.md with issue links (#363-#372) - Updated github-issues-fsharp-backend.md with actual issue numbers - Fixed duplicate metadata fields in issue templates - All child issues now properly linked to Epic #363 - Epic #363 updated with comprehensive description and child issue list Related: #363, #364, #365, #366, #367, #368, #369, #370, #371, #372 🤖 Generated with [Claude Code](https://claude.com/claude-code)

Phase 7: Testing and Documentation #372

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions