BoundaryML
baml
Blog
Docs
Changelog
Blog
Docs
Changelog
Overview
Branches
Benchmarks
Runs
Performance History
Latest Results
fix(go): handle optional image encoding (*types.Image) (#2996) ## Summary - Fix Go encoder to handle pointer-to-interface types like `*types.Image` - In Go, `*types.Image` (pointer to interface) doesn't implement `InternalBamlSerializer`, but the dereferenced value does - Also fixes a bug where all media types were converted to `Image` in streaming type conversion ## Test plan - [x] All 115 type_gen tests pass across Go, Python, TypeScript, Rust - [x] Manual verification of optional image encoding - [x] Added Go integration tests `TestOptionalImageInputNil` and `TestOptionalImageInputWithValue` š¤ Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
canary
16 minutes ago
Fix Go codegen for pure-dynamic classes (@@dynamic with no static fields) (#2995) ## Summary - Fixed Go code generator producing incorrect `Decode()` methods for classes marked with `@@dynamic` that have **no static fields** - The generated code was initializing `DynamicProperties` but never populating it because the field iteration loop was conditionally skipped - Root cause: In `class.go.j2`, the condition `{% if fields.len() > 0 %}` skipped the loop when there were no static fields - Fix: Changed condition to `{% if !fields.is_empty() || dynamic %}` so the loop runs for dynamic classes ## Test plan - [x] Added `PureDynamic` test class to dynamic_types test schema (class with `@@dynamic` but no static fields) - [x] Added Go test `TestPureDynamicClassE2E` to verify dynamic properties are populated - [x] Added Rust tests for pure-dynamic class handling - [x] Verified generated `PureDynamic.Decode()` now includes the field iteration loop - [ ] Run full E2E tests with API key: `RUN_GENERATOR_TESTS=1 infisical run --env=test -- cargo test -- dynamic_types_evaluate` š¤ Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
canary
25 minutes ago
feat(rig): add generation headers to all template-generated files All files generated from crate_templates now include a header comment identifying their source. This makes it clear which files are generated and helps prevent accidental manual edits. Changes: - Cargo.toml, build.rs, src/lib.rs: Add "# Generated by baml_tools_rig from crate_templates/{language}" header - test.sh, test.ps1: Add same header (line 2 for shell scripts with shebang) - baml_tools_rig validation: Check all files have correct headers - Validates first line for .toml and .rs files - Validates embedded script strings in build.rs contain headers This ensures generated files are clearly marked and validated. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
feat/python-codegen-e2e-rig
40 minutes ago
feat(rig): add generation headers to all template-generated files All files generated from crate_templates now include a header comment identifying their source. This makes it clear which files are generated and helps prevent accidental manual edits. Changes: - Cargo.toml, build.rs, src/lib.rs: Add "# Generated by baml_tools_rig from crate_templates/{language}" header - test.sh, test.ps1: Add same header (line 2 for shell scripts with shebang) - baml_tools_rig validation: Check all files have correct headers - Validates first line for .toml and .rs files - Validates embedded script strings in build.rs contain headers This ensures generated files are clearly marked and validated. Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
feat/python-codegen-e2e-rig
53 minutes ago
feat(rig): add PowerShell test script support for Windows Problem: Rig tests were failing on Windows CI because they tried to run bash scripts (test.sh) but Windows doesn't have WSL installed. Solution: Generate both test.sh (Unix) and test.ps1 (Windows) scripts, and update the Rust tests to use the appropriate script based on platform. Changes: - build.rs.template: Generate both test.sh and test.ps1 - lib.rs.template: Use platform-specific script (#[cfg(windows)]) - test.ps1: PowerShell equivalent with same checks (syntax, ruff, pytest) - Platform detection: Uses "powershell -File" on Windows, "bash" on Unix Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
feat/python-codegen-e2e-rig
1 hour ago
fix(rig): add rustfmt to toolchain and improve diff diagnostics Problem: rig --check was failing in CI but passing locally due to: 1. rustfmt not installed in CI (not in rust-toolchain.toml components) 2. format_rust_code() silently returning unformatted code on failure 3. Unhelpful error messages (just "DIFF: crate_name") Changes: - Add rustfmt to rust-toolchain.toml components - Improve diff error messages to show which file differs and size differences - Add stderr logging when rustfmt fails - Rename has_differences -> check_differences, return Option<String> with details Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
feat/python-codegen-e2e-rig
1 hour ago
fix(ci): install uv and exclude rig tests from insta/wasm - Install uv before nextest runs on all platforms (Linux, macOS, Windows) - Exclude rig_python_* crates from insta snapshot tests (no snapshots in rig tests) - Exclude rig_python_* crates from wasm builds and clippy (not wasm-compatible) - Use PowerShell installer for uv on Windows Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
feat/python-codegen-e2e-rig
2 hours ago
refactor: move rig check to .pre-commit-config.yaml - Add rig hook to .pre-commit-config.yaml - Remove scripts/pre-commit and scripts/README.md (superseded by prek) - Consolidate all pre-commit checks in one place
feat/python-codegen-e2e-rig
2 hours ago
Active Branches
feat: Python codegen and E2E rig test framework
last run
40 minutes ago
#2994
CodSpeed Performance Gauge
-1%
progress
last run
10 hours ago
#2990
CodSpeed Performance Gauge
+74%
Sam/rp2
last run
2 days ago
#2963
CodSpeed Performance Gauge
+99%
Ā© 2026 CodSpeed Technology
Home
Terms
Privacy
Docs