Fix template argument resolution for complex typedef chains (e.g.
vector<T>::reference through allocator_traits) by:
- Adding ownership checking in find_argument to skip frames from
unrelated templates at the same depth
- Tracking entry stack size in TransformDependentNameType to prevent
frame leaks from nested NNS qualifier resolutions
- Extracting template info directly from concrete TSTs in lookup to
avoid stack corruption
- Adding push_nns_qualifier_frames for multi-level dependent chains
Also removes debug traces from TemplateResolver.Standard test.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
When rerunning failed validate jobs after pushing fixes, the checkout
now fetches the branch HEAD instead of the original trigger commit.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add clangOptions to Debug library list (split from clangDriver in 22.1).
Update test expectations for ElaboratedType removal and NNS changes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Restructure build-llvm.yml so that LLVM build and clice build/test
run in separate jobs. This allows "rerun failed jobs" to skip the
expensive LLVM rebuild when only clice code needs fixing.
- Rename build → build-llvm (LLVM only, no clice build/test/prune)
- Add validate job that reuses test-cmake.yml with llvm_from_artifact
- Upload raw LLVM install directory as artifact for validation
- Keep .tar.xz packaging for clice-llvm release
- Fix tar to use flat directory structure (no build-install wrapper)
- Filter download-llvm.sh to only fetch .tar.xz artifacts
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Major LLVM 22.1 breaking changes addressed:
- NestedNameSpecifier changed from pointer to value type
- ElaboratedType removed (merged into individual type nodes)
- DependentTemplateSpecializationType merged into TemplateSpecializationType
- TreeTransform overrides now require additional parameters
- TagDecl::getTypeForDecl() removed, use ASTContext::getTagType()
- InjectedClassNameType::getInjectedSpecializationType() removed
- TypedefTypeLoc::getTypedefNameDecl() renamed to getDecl()
- AnonymousTagLocations renamed to AnonymousTagNameStyle
- DIAG macro expanded from 11 to 13 parameters
- Header moves: clang/Driver/Options.h → clang/Options/Options.h
- ClangTidyModuleRegistry.h renamed to ClangTidyModule.h
- createDiagnostics/createFileManager API changes
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
LLVM 22.1 introduced new CMake targets (clangAnalysisLifetimeSafety,
clangOptions, clangDependencyScanning, LLVMPlugins) that existing
components depend on. Add them to the distribution component list.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- **`[[rules]]`**: TOML array-of-tables config for per-file compilation
flag rules with glob pattern matching (`append`/`remove`). Patterns are
pre-compiled at config load time. Rules whose patterns all fail to
compile are dropped entirely (no silent no-op entries), and rules now
apply uniformly to every compilation — including the header-context
fallback path used when editing a header without its own CDB entry.
- **CDB auto-scan**: Default search scans workspace root + all immediate
subdirectories for `compile_commands.json`, replacing the hardcoded
directory list.
- **LSP `initializationOptions`**: Clients can pass config as JSON via
the LSP initialize request; priority is `initializationOptions >
clice.toml > defaults`.
- **XDG cache paths**: Default cache/index/logging paths prefer
`$XDG_CACHE_HOME/clice/<workspace-hash>/`; fall back to
`$HOME/.cache/clice/<hash>/`, then `<workspace>/.clice/`.
- **`${workspace}` substitution**: supported in `cache_dir`,
`index_dir`, `logging_dir`, and every `compile_commands_paths` entry.
No-op when `workspace_root` is empty.
- **Partial config support**: All TOML/JSON fields are optional via
`kota::meta::defaulted<T>`, so minimal config files work correctly.
- **Detailed diagnostics**: malformed `clice.toml` now logs line, column
and parser description (via toml++ direct parse); a malformed workspace
config surfaces a clear fallback warning instead of silently reverting
to defaults.
## Test plan
- [x] 28 unit tests for config (full suite 545 unit tests pass, Debug)
- [x] 119 integration tests pass
- [x] 2 smoke tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* XDG-based, workspace-scoped project cache (PCH/PCM and header caches
moved under project cache) with workspace fallback
* Initialization options JSON can override config (takes precedence over
file/defaults)
* Per-file pattern rules to append/remove compile flags; expanded
discovery of compilation databases (multiple paths)
* **Refactor**
* Configuration fields reorganized under a project scope; runtime
behavior now respects project-scoped values
* **Tests**
* New unit and integration tests for config parsing, rule matching, and
persistent cache behavior
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Check `CXAvailability_Deprecated` on `CodeCompletionResult` and set
`CompletionItemTag::Deprecated`
- Editors render deprecated completions with a strikethrough on the
label
## Test plan
- [x] `DeprecatedTag` — `[[deprecated]]` function gets the tag
- [x] `NotDeprecated` — normal function has no Deprecated tag
- [x] All 491 unit tests pass
- [x] `pixi run format` clean
Stacked on #411.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Code completion now marks deprecated declarations with a deprecated
tag so users can see deprecated items in completion lists.
* **Tests**
* Added unit tests ensuring deprecated declarations produce completion
items with the deprecated tag and non-deprecated items do not.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
This PR adds cross-compilation support for three new target platforms,
upgrades LLVM to 21.1.8, and overhauls the CI pipelines around
cross-builds and testing.
## Cross-compilation
New target triples accepted via `-DCLICE_TARGET_TRIPLE=...`:
| Target triple | Host | Output |
|---|---|---|
| `x86_64-apple-darwin` | macos-15 (arm64) | macOS x64 |
| `aarch64-linux-gnu` | ubuntu-24.04 (x64) | Linux arm64 |
| `aarch64-pc-windows-msvc` | windows-2025 (x64) | Windows arm64 |
- `cmake/toolchain.cmake` — maps `CLICE_TARGET_TRIPLE` to
`CMAKE_SYSTEM_NAME`/`CMAKE_SYSTEM_PROCESSOR`/compiler `--target`; picks
up conda aarch64 sysroot when cross-compiling Linux.
- `cmake/llvm.cmake` — forwards target platform/arch to `setup-llvm.py`
so the right prebuilt LLVM is downloaded for the target.
- `CMakeLists.txt` — uses a host-side `flatc` from `PATH` under
`CMAKE_CROSSCOMPILING` instead of the in-tree target build.
- `pixi.toml`:
- Adds `osx-64`, `linux-aarch64`, `win-arm64` platforms.
- New environments: `cross-macos-x64`, `cross-linux-aarch64` (adds
`gcc_linux-aarch64` + `sysroot_linux-aarch64`), `cross-windows-arm64`.
- New lightweight `test-run` env used on native ARM/x64 runners to
execute cross-built artifacts (pulls in upstream clang+lld on macOS so
tests don't fall back to Apple clang).
- `scripts/activate_cross_linux.sh` — exports `CONDA_PREFIX`-relative
paths for the aarch64 toolchain.
- `scripts/build-llvm.py` — `--target-triple` support and a
`build_native_tools()` helper that produces host `llvm-tblgen` /
`clang-tblgen` needed when cross-compiling LLVM itself.
## LLVM upgrade 21.1.4 → 21.1.8
- `cmake/package.cmake` bumps `setup_llvm("21.1.8")`.
- `config/llvm-manifest.json` regenerated with 6 new cross-compiled
entries and a new `arch` field on every entry so lookup is `(version,
platform, arch, lto, build_type)`.
- `scripts/setup-llvm.py` — honours the new `arch` field when resolving
artifacts.
- `scripts/update-llvm-version.py` (new) — single-call version bump
across `package.cmake` + manifest.
- `scripts/validate-llvm-components.py` (new) — scans the LLVM source
tree for library targets and diffs them against
`scripts/llvm-components.json` to catch stale/misspelled component names
before a build.
- `scripts/llvm-components.json` (new) — explicit allow-list of required
LLVM/Clang library targets used by `build-llvm.py`.
## CI changes
- `.github/workflows/build-llvm.yml`:
- Adds `workflow_dispatch` with `llvm_version`, `skip_upload`, `skip_pr`
inputs.
- Matrix extended with the 6 cross-compile entries (2 per new platform:
RelWithDebInfo ± LTO).
- `build clice` / test / prune steps gated on `!matrix.target_triple`
for cross-builds; cross-built LTO entries apply the native prune
manifest (arch-independent).
- Cross-compiled binary architecture is verified with `file(1)`.
- New `upload` job triggered by `workflow_dispatch` pushes artifacts to
`clice-io/clice-llvm` and hands the manifest off to the next job.
- `.github/workflows/test-cmake.yml`:
- Build matrix gains three `build_only: true` cross entries that upload
`bin/` + `lib/` artifacts.
- New `test-cross` job runs on native `macos-15-intel`,
`ubuntu-24.04-arm`, `windows-11-arm` runners, downloads the cross-built
artifacts, and runs unit / integration / smoke tests under the
`test-run` pixi env.
- Cache keys now include `target_triple` so native and cross builds
don't collide.
- `.github/workflows/publish-clice.yml`:
- Three additional release artifacts for the new targets
(`clice-x86_64-macos-darwin`, `clice-aarch64-linux-gnu`,
`clice-aarch64-windows-msvc`), each with a matching `-symbol` archive.
## Compatibility
- All existing native builds and tests are preserved; cross entries are
additive.
- `Debug` + ASAN remains disabled on Windows (`llvm_mode == Debug && os
== windows-*` no longer appends `-asan`).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- The `eventide` dep was renamed to
[kotatsu](https://github.com/clice-io/kotatsu) with a broad rename of
CMake identifiers, namespaces, header paths, and a few module reorgs
(`serde` → `codec`, `reflection` → `meta`, `common` → `support`). Align
clice to the new names.
- CMake: FetchContent target, option prefix (`ETD_*` → `KOTA_*`,
`ETD_SERDE_*` → `KOTA_CODEC_*`), target names
(`eventide::{ipc::lsp,serde::toml,deco,zest}` →
`kota::{ipc::lsp,codec::toml,deco,zest}`).
- Namespaces: `eventide::` → `kota::`, `eventide::serde::` →
`kota::codec::`, `eventide::refl::` → `kota::meta::`. The short `et`
alias is dropped — all usages now spell `kota::` directly.
- Headers: `eventide/*` → `kota/*`, including special cases
`serde/serde/raw_value.h` → `codec/raw_value.h`, `ipc/json_codec.h` →
`ipc/codec/json.h`, `common/meta.h` → `support/type_traits.h`,
`common/ranges.h` → `support/ranges.h`.
- Kotatsu split `JsonPeer` / `BincodePeer` out of `ipc/peer.h` into the
codec-specific headers; added `kota/ipc/codec/{json,bincode}.h` includes
where those types are used.
- Depends on clice-io/kotatsu#110 (already merged) to prevent `-Wall
-Wextra -Werror` from transitively propagating out of
`kota::project_options`.
## Test plan
- [x] `pixi run unit-test RelWithDebInfo` — 518/518 pass (9 skipped,
unchanged from main)
- [x] `pixi run integration-test RelWithDebInfo` — 119/119 pass
- [x] `pixi run smoke-test RelWithDebInfo` — 2/2 pass
- [x] `pixi run format` clean
## Notes
- `tests/smoke/rapid_edit.jsonl` was intentionally left untouched: the
embedded `#include "eventide/..."` strings are frozen snapshots of file
contents the client sent at record time, not clice source.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Updated internal dependencies from `eventide` to `kota`, including
async runtime, IPC transport, serialization codec, and metadata
libraries.
* Updated build configuration and CMake variables to align with the new
dependency.
* **Refactor**
* Migrated internal implementation to use `kota` namespace and APIs
throughout the codebase.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
## Summary
- Replace hand-written character scanning in `document_links.cpp` with
the project's `Lexer` class for finding filename arguments in
preprocessor directives
- Extend `Lexer` to activate `header_name` mode for
`#embed`/`#include_next`, and expose `set_header_name_mode()` for
`__has_include`/`__has_embed` contexts
- Remove unused `Include::filename_range` field (had a latent assert
crash on macro-expanded includes)
- Add `MacroInclude` unit test covering `#include MACRO` scenario
## Test plan
- [x] 498 unit tests pass (including new `MacroInclude` test)
- [x] 119 integration tests pass
- [x] 2/2 smoke tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Document links now resolve includes written via macros; directive
parsing recognizes include, include_next, embed and __has_* patterns
more reliably using lexer-driven argument detection.
* **Refactor**
* Removed an internal filename-range field previously stored for include
directives.
* **Tests**
* Added unit tests covering directive argument extraction and
macro-based include linking.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Add `add_module()` and `compile_with_modules()` to the `Tester` test
framework
- Supports both separate `add_module()` calls and single-string
`#[filename]` syntax via `add_files()`
- Automatically scans module dependencies with `scan_precise`,
topologically sorts, builds PCMs in order, then compiles the main file
- Temporary PCM files cleaned up automatically in destructor
- Migrated `ModuleImport` and `ModuleReexport` semantic tokens tests to
use the new API
## Test plan
- [x] All 505 unit tests pass
- [x] All 113 integration tests pass
- [x] All 2 smoke tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Tests**
* Centralized, module-aware test compilation with automatic module
discovery, dependency ordering, and cycle detection.
* Unified "compile with modules" flow; tests now add module sources
directly and no longer manage temporary module artifacts manually.
* Reduced duplicated compile/diagnostic logic and improved cleanup of
generated artifacts.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- PCH compilation now serializes document links via `pch_links_json` in
`BuildResult` and stores them in `PCHState`
- Master server merges PCH document links with main-file links on
`textDocument/documentLink` requests, fixing missing links for
`#include` directives inside the preamble
- Adds document link support for `#embed` and `__has_embed` directives
## Test plan
- [x] Unit tests: `DocumentLink.Embed` and `DocumentLink.HasEmbed` added
- [x] Integration tests: `test_document_links.py` verifies PCH + main
merge and `#embed` links
- [x] All 483 unit tests pass
- [x] All 4 integration tests pass
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Document links now detect embeds and __has_embed directives for both
quoted and angled filenames.
* Document links produced during precompiled builds are cached and
merged into document-link responses for more complete link sets.
* **Tests**
* Added integration tests for merged PCH/main links and embed/has-embed
cases.
* Added unit tests verifying embed handling under C++23.
* **Chores**
* Added test fixtures and compile command entries for document-links
tests.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Highlight module name identifiers (e.g. `foo`, `bar` in `export module
foo.bar;`) as `SymbolKind::Module` in semantic tokens
- Highlight import module names (e.g. `foo` in `import foo;`) using
`directives.imports` name locations
- Module declarations use `getCurrentNamedModule()->DefinitionLoc` +
lexer scan to find name tokens
## Test plan
- [x] `SemanticTokens.ModuleDeclaration` — `export module foo;`
- [x] `SemanticTokens.ModuleDeclarationDotted` — `export module
foo.bar;`
- [x] `SemanticTokens.ModuleImport` — PCM build + `import foo;`
- [x] All 16 SemanticTokens tests pass, no regressions
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Enhanced semantic token support for C++20 modules, including dotted
module names, partitions, fragments, imports and re-exports for more
accurate highlighting.
* **Bug Fixes**
* Improved conflict resolution so directive tokens no longer mask other
semantic kinds; ensures `module`/`import` used as identifiers are
tokenized correctly.
* **Tests**
* Added unit tests covering module declarations, imports, partitions and
edge cases.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Fixes a bug where constructors and deduction guides had labels like
`vector<_Tp, _Alloc>` instead of just `vector`, causing:
1. Label deduplication to fail (class `vector` != constructor
`vector<_Tp, _Alloc>`)
2. Selecting the completion to insert invalid text `vector<_Tp, _Alloc>`
Now uses `getParent()->getName()` for constructors and
`getDeducedTemplate()->getName()` for deduction guides.
## Test plan
- [x] All 494 unit tests pass (existing `DeduplicateByLabel` test covers
this)
- [x] `pixi run format` clean
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Bug Fixes**
* Constructor and deduction-guide completions now show the parent type
name without template parameters, improving readability and preventing
duplicate entries.
* **Tests**
* Added a unit test verifying completion items for these entries use the
parent type name (no template-parameterized labels) and insertion text
starts with that name.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Add `.claude/CLAUDE.md` with project guide, C++ coding style,
correction patterns from past AI interactions, and pre-commit/pre-PR
workflows
- Add `.claude/commands/` with `/build`, `/test`, `/format` slash
commands for Claude Code
- Add `.clang-tidy` configuration
- Add `scripts/run_clang_tidy.py` for parallel clang-tidy with progress
reporting (reads CDB, filters to `src/` and `tests/` only)
- Add pixi tasks: `clang-tidy`, `lint-cpp`, `lint-python`, `lint`
- Add `clang-tools` dependency for clang-tidy
- Update `.gitignore` to track `.claude/CLAUDE.md` and
`.claude/commands/`
## Test plan
- [x] `pixi run format` passes
- [x] `pixi run lint-python` passes (ruff check)
- [x] `scripts/run_clang_tidy.py` correctly filters to project files
only
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Generate LSP snippet placeholders (`${1:param}`, `${2:param}`) for
function and method completions in non-bundle mode
- Controlled by
`CodeCompletionOptions::enable_function_arguments_snippet` (default off)
- No-arg functions produce plain text insertion (no empty snippet)
- Bundle mode is unaffected — snippets only apply when each overload is
a separate item
- Optional chunks (default arguments) are skipped in snippet generation
## Example
```
// Before: typing "fo" and selecting foooo inserts just "foooo"
// After: typing "fo" and selecting foooo inserts "foooo(${1:int x}, ${2:float y})"
```
## Test plan
- [x] `SnippetFunctionArgs` — verifies placeholders are generated
- [x] `SnippetNoArgs` — no-arg functions don't produce snippet
- [x] `SnippetDisabled` — respects the option flag
- [x] `SnippetBundleMode` — bundle mode doesn't generate snippets
- [x] `SnippetMethod` — works for member methods too
- [x] All 494 unit tests pass
- [x] `pixi run format` clean
Stacked on #411.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Release Notes
* **New Features**
* Code completion now generates function argument snippets with
interactive placeholders, helping users efficiently navigate through
parameters during autocompletion. The feature works with functions and
methods, with configurable options to control behavior for overloaded
scenarios.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Extract function/method signatures from Clang `CodeCompletionString`
into `labelDetails.detail` (parameter list) and
`labelDetails.description` (return type)
- Filter `_`/`__` prefixed internal symbols (e.g. `_Vector_base`,
`_Alloc`) unless the user explicitly typed `_`
- Fix `completion_kind` isa ordering: `CXXMethodDecl` checked before
`FunctionDecl` so methods get correct Kind
- Bundle mode: extend overload bundling to Method and Constructor (was
Function only)
- Bundle mode: deduplicate by label — when the same name appears as
Class + Constructor + deduction guide, keep only one (priority: Class >
Function > Constructor)
- Bundled overloads show `(…) +N overloads` in `labelDetails.detail`
instead of `detail`
## Test plan
- [x] 12 unit tests covering: signature extraction, return type,
overload bundling, underscore filtering, label deduplication, non-bundle
mode, method signatures
- [x] All 489 unit tests pass
- [x] `pixi run format` applied
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Release Notes
* **New Features**
* Code completion now displays function signatures and return types in
completion items
* Overloaded functions are bundled together with a count indicator
* Internal symbols (underscore-prefixed) are filtered from suggestions
unless explicitly typed
* Duplicate completion items are deduplicated while preserving
higher-priority variants
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Extract shared test utilities into `tests/integration/utils/` (client,
workspace, assertions, wait, cache)
- Migrate 12 test files into categorized subdirectories: `lifecycle/`,
`compilation/`, `features/`, `modules/`, `extensions/`, `stress/`
- Merge `test_include_completion.py` + `test_import_completion.py` →
`features/test_completion.py`
- Remove stale directory-tree comments and section divider comments
## Test plan
- [x] `pytest --collect-only` collects all 113 tests
- [x] All test module imports verified
- [x] `pixi run format` applied
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Tests**
* Reorganized integration suites: added new feature tests (completion,
server) and removed older duplicated modules.
* Centralized shared test utilities and assertion helpers for
diagnostics, workspace operations, waiting/synchronization, and cache
inspection.
* **Chores / Refactor**
* Standardized test client lifecycle and helper usage across suites for
more reliable test flows.
* Improved server session lifecycle handling for more predictable
document/session resets during tests.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- Replace flat `CompilationContext { directory, arguments }` with a
three-layer abstraction: `ResolvedFlags` (file-independent flags) →
`CompileCommand` (+ source file) → `to_argv()` (full argv on demand)
- `ResolvedFlags.flags` never contains source file path or
`-main-file-name`, making it directly usable as a clean cache key input
(e.g. PCH sharing across files with identical preambles)
- `to_argv()` handles `-main-file-name` insertion for cc1 mode
automatically — consumers no longer need to search/replace in the
argument list
- Eliminates the pollute-then-clean anti-pattern in `lookup()` and the
manual source-file replacement in `fill_header_context_args()`
## Test plan
- [x] `pixi run format` — no changes
- [x] `pixi run unit-test` — 481 passed
- [x] `pixi run integration-test` — 113 passed
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Refactor**
* Unified compile-command handling across the server and tools for more
consistent argument and flag behavior (driver vs frontend modes).
* **New Features**
* Added an LRU-backed in-memory cache to improve performance and
eviction control.
* **Chores**
* Added an option to control injection of resource-directory flags
(enabled by default).
* **Tests**
* Updated unit and integration tests to adopt the new command
representation and verify cache behavior.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
## Summary
- **Extract `Indexer` class** — owns all index state (ProjectIndex,
MergedIndex shards, OpenFileIndex) and query methods (definition,
references, call/type hierarchy, workspace symbol search)
- **Extract `Compiler` class** — owns document state, PCH/PCM cache,
compile argument resolution, header context, `ensure_compiled`, and
worker forwarding
- **MasterServer is now a pure LSP handler registration layer** (~700
lines, down from ~3200)
- **`MergedIndexShard`** wraps `index::MergedIndex` with a lazily-cached
PositionMapper; `OpenFileIndex` gains matching
`find_occurrence()`/`find_relations()` APIs — callers get pre-converted
LSP ranges directly
- **Indexer returns typed values** (`vector<Location>`,
`vector<CallHierarchyIncomingCall>`, etc.) instead of pre-serialized
JSON, fixing the references handler from JSON string surgery to simple
vector concatenation
- **Fix**: duplicate `workspace/symbol` loop in the original code
## Test plan
- [x] 465 unit tests pass
- [x] 113 integration tests pass
- [x] 2/2 smoke tests pass
- [x] `clang-format` applied
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Server-side C++ compilation orchestration (module & precompiled header
builds) with LSP-integrated document handling.
* **Improvements**
* Deterministic, persistent, dependency-aware caching to avoid redundant
rebuilds and speed up incremental work.
* Better cross-file indexing and navigation, improved diagnostics and
more reliable include/import-aware completions.
* **Tests**
* Unit tests updated to the unified worker query/build request shapes.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Piggyback index construction onto existing compilation steps,
eliminating redundant recompilation in background indexing:
- **`TUIndex::build` gains `interested_only` parameter**: `true`
traverses only the main file's top-level decls; `false` (default)
traverses the full AST
- **PCH build indexes preamble headers**: stateless worker calls
`TUIndex::build(unit)` (full traversal) after successful `BuildPCH`,
clears `main_file_index`, serializes and sends back; master merges into
MergedIndex
- **PCM build indexes module interface**: stateless worker calls
`TUIndex::build(unit, true)` after successful `BuildPCM`; master merges
into MergedIndex
- **Open-file compile indexes main file**: stateful worker calls
`TUIndex::build(unit, true)` after successful `Compile`, serialized in
`CompileResult`
- **New `OpenFileIndex` in-memory structure**: master holds `FileIndex +
SymbolTable + buffer text` per open file — not persisted to disk, not
merged, discarded on close
- **Dual-source query path**: `query_index_relations`,
`lookup_symbol_at_position`, `find_symbol_definition_location`, all
hierarchy handlers, and `workspace/symbol` check `OpenFileIndex` first
(fresher), then fall back to `MergedIndex` (disk-indexed)
- **Background indexing skips open files**: checked via
`documents.count()`; on `didClose` the file is re-queued into
`index_queue`
- **`didSave` re-queues non-open dependents**: dirtied files from
`compile_graph->update()` that are not open get pushed into
`index_queue` for background re-indexing
- **Extract `lookup_occurrence` helper**: binary search + forward scan
picking the innermost (narrowest) match, replacing a broken
`while/break/break` pattern
- **Extract `find_symbol_info` helper**: consolidates 6 duplicated
"search open file indices then ProjectIndex" lookups into one method
- **`resolve_hierarchy_item` checks open file indices**: no longer
limited to ProjectIndex only
## Test plan
- [x] 465 unit tests pass
- [x] 105 integration tests pass (including all `test_index` cases:
GoToDefinition, FindReferences, CallHierarchy, TypeHierarchy,
WorkspaceSymbol)
- [x] Manual: open a file and immediately use GoToDefinition — should
work without waiting for background indexing
- [x] Manual: close a file and verify background indexing picks it up
and produces a MergedIndex shard
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Add three LSP protocol extensions that allow users to manage compilation
contexts for header files and source files with multiple CDB entries.
### Protocol extensions (`protocol.h`)
| Command | Purpose |
|---------|---------|
| `clice/queryContext` | List all possible contexts for a file. Headers
→ host source files; sources → CDB entries. Paginated (10 per page,
`offset` param). |
| `clice/currentContext` | Query the active context override for a file
(null if default). |
| `clice/switchContext` | Set the active context, invalidate caches,
trigger recompilation. |
### Header context resolution (`master_server.cpp`,
`dependency_graph.cpp`)
- `find_host_sources()`: BFS the reverse include graph to find source
files that transitively include a header
- `find_include_chain()`: BFS the forward include graph to find the
shortest include chain from host to header
- `resolve_header_context()`: walks the include chain, extracts content
before each `#include` directive, concatenates with `#line` markers into
a preamble file (hash-addressed under `.clice/header_context/`)
- `fill_header_context_args()`: uses the host source's CDB entry,
replaces source path with header path, injects `-include preamble.h`
### Compilation flow
- Default: headers compile as standalone files (no context)
- After `switchContext`: `fill_compile_args` checks `active_contexts`
first → uses host's CDB entry + preamble injection
- Fallback: if no CDB entry and no active context, auto-resolves via
`resolve_header_context`
- `#include` directive matching uses precise filename extraction from
`"..."` / `<...>`, not substring matching
### Source file multiple contexts (`multi_context` workspace)
- `queryContext` on a source file returns all CDB entries with
distinguishing labels (extracted from `-D`, `-O`, `-std=` flags)
### Test data
- `header_context/`: non-self-contained 3-level chain (`main.cpp` →
`utils.h` → `inner.h`), `types.h` provides `Point` struct
- `multi_context/`: single source with two CDB entries (`-DCONFIG_A`,
`-DCONFIG_B`)
### Tests (9 integration tests)
- queryContext returns host sources for headers
- queryContext returns CDB entries for source files
- currentContext defaults to null
- switchContext sets active context, currentContext reflects it
- Full flow: open → query → switch → hover works in non-self-contained
header
- Deep nested: switchContext + hover on `inner.h` (3 levels deep)
- Multiple CDB entries: queryContext returns both CONFIG_A and CONFIG_B
## Test plan
- [x] Unit tests: 465 passed
- [x] Integration tests: 113 passed (9 new header context tests)
- [x] Smoke test: 1/1 passed
- [ ] Manual VSCode testing
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This patch restricts the GLOB expression to only LLVM & Clang libraries,
avoiding linking to a large list of libraries when using system LLVM.
Also, this patch explicitly avoids linking to `LLVM-<ver>` and
`clang-cpp`, as they will cause lots of duplicate symbol errors
Below is a small set of errors generated when linking to `clang-cpp`:
```
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::DynTypedMatcher::constructRestrictedWrapper(clang::ast_matchers::internal::DynTypedMatcher const&, clang::ASTNodeKind)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::Stmt::getEndLoc() const
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::createAndVerifyRegex(llvm::StringRef, llvm::Regex::RegexFlags, llvm::StringRef)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::BoundNodesTreeBuilder::addMatch(clang::ast_matchers::internal::BoundNodesTreeBuilder const&)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::Stmt::getStmtClassName() const
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::getExpansionLocOfMacro(llvm::StringRef, clang::SourceLocation, clang::ASTContext const&)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::transformer::remove(std::__1::function<llvm::Expected<clang::CharSourceRange> (clang::ast_matchers::MatchFinder::MatchResult const&)>)
>>> defined at libclangTransformer.a(RewriteRule.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::hasAnyOverloadedOperatorNameFunc(llvm::ArrayRef<llvm::StringRef const*>)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::transformer::detail::makeEditGenerator(clang::transformer::ASTEdit)
>>> defined at libclangTransformer.a(RewriteRule.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::Stmt::stripLabelLikeStatements() const
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::Stmt::IgnoreContainers(bool)
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::WhileStmt::getConditionVariable()
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ValueStmt::getExprStmt() const
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::ast_matchers::internal::matchesAnyBase(clang::CXXRecordDecl const&, clang::ast_matchers::internal::Matcher<clang::CXXBaseSpecifier> const&, clang::ast_matchers::internal::ASTMatchFinder*, clang::ast_matchers::internal::BoundNodesTreeBuilder*)
>>> defined at libclangASTMatchers.a(ASTMatchersInternal.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::IfStmt::getConditionVariable()
>>> defined at libclangAST.a(Stmt.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
ld.lld: error: duplicate symbol: clang::transformer::detail::buildMatchers(clang::transformer::RewriteRuleBase const&)
>>> defined at libclangTransformer.a(RewriteRule.cpp.obj)
>>> defined at libclang-cpp.dll.a(libclang-cpp.dll)
```
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Improved build configuration to explicitly discover and link LLVM and
Clang libraries plus optional z/zstd compression libraries; updated
linking to include these groups and retained the static Clang build
definition, resulting in more reliable and consistent compilation across
environments.
<sub>✏️ Tip: You can customize this high-level summary in your review
settings.</sub>
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Summary
- Fill in Visual Studio, Makefile, Meson sections (previously TODO)
- Expand Xmake section with CLI and VSCode extension workflows
- Simplify Others section to recommend
[catter](https://github.com/clice-io/catter)
- Fix CJK-Latin spacing in Chinese docs
- English and Chinese docs updated in sync
Supersedes #313 by @Stehsaer.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Stehsaer <Stehsaer@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Add `Embed` and `HasEmbed` structs to `directive.h` for storing
`#embed` / `__has_embed` directive info
- Implement `EmbedDirective` and `HasEmbed` PPCallbacks in
`DirectiveCollector`, using the current `CompilationUnitRef` API
- Add unit tests for both directives (including non-existent file
handling)
Rebased onto current main, resolving conflicts from the `Compiler/` →
`compile/` restructuring and the `CompilationUnitRef` API migration.
Original PR by @Guo-Shiyu.
## Test plan
- [x] `pixi run unit-test` — 465 tests passed, 0 failures
Co-authored-by: ykiko <ykikoykikoykiko@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Disable CodeRabbit's auto-generated PR summary while keeping code
review enabled
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Configuration updates to automated review processes.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- Add `conventional-commit` CI job that validates PR titles and push
commit messages follow the `type(scope)?: description` format
- Valid types: `feat`, `fix`, `refactor`, `chore`, `build`, `ci`,
`docs`, `test`, `perf`, `style`, `revert`
- Runs on both PR and push events, skips tag pushes
- Zero dependencies, pure bash regex check
## Test plan
- [x] PR title of this PR itself passes the check
- [ ] CI runs and passes
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Enhanced continuous integration pipeline to validate commit message
formatting standards on all pull requests and commits.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- add declaration helpers to classify semantic token modifiers such as
readonly, static, abstract, virtual, default library, and
constructor/destructor
- unwrap template declarations before applying attribute-style modifiers
so modifier checks hit the underlying declaration
- keep templated/dependent-name handling and expand emitted semantic
token modifiers in `semantic_tokens.cpp`
## Testing
- Not run (not requested in this turn)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Enhanced semantic token analysis to improve code recognition and more
accurate classification of declarations across C++ and Objective-C
codebases.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Summary
- register workspace and text document capabilities through the
structured LSP capability types
- advertise completion, signature help, declaration, definition,
implementation, type definition, and reference support more explicitly
- add placeholder handlers for declaration, type definition, and
implementation requests so the advertised capabilities have matching
routes
## Testing
- Not run (not requested)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added workspace folder support for improved project tracking.
* Registered navigation handlers for type-definition, implementation,
and declaration (currently return a “not supported yet” placeholder).
* **Improvements**
* Enhanced completion and signature help with explicit trigger
characters and clearer capability declarations.
* **Tests**
* Relaxed capability assertions to recognize more nuanced
enabled/disabled states.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Summary
- classify unresolved using declarations with the `dependentName`
semantic token modifier
- include the declaration modifier on definitions to match clangd's
semantic token behavior
- extend the semantic token modifier legend coverage in integration
tests
## Testing
- pixi run python -m pytest -s --log-cli-level=INFO
tests/integration/test_server.py -k 'semantic_token_modifier_legend or
capabilities' --executable=./build/bin/clice
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Added many new semantic token modifiers (deprecated, deduced,
readonly, static, abstract, virtual, dependent-name,
constructor/destructor, user-defined, mutable-usage flags) and new scope
markers (function, class, file, global).
* Improved tagging for declarations, definitions and dependent names in
semantic tokens.
* **Tests**
* Added an integration test verifying the semantic token modifier legend
and its ordering.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
## Summary
Implement structured file-based logging with per-component separation
and crash stacktrace capture.
### Log output structure
```
.clice/logs/2026-04-05_10-30-00_<pid>/
master.log
SF-0.log
SF-1.log
SL-0.log
```
### Changes
**Logging infrastructure** (`logging.h`, `logging.cpp`)
- `file_logger()` creates a dual-sink logger (file + stderr), so logs go
to both the file and terminal
- Pre-checks log directory creation and file writability before
constructing spdlog sinks; falls back to existing stderr logger on
failure
- `install_crash_handler()` uses LLVM's `AddSignalHandler` +
`PrintStackTraceOnErrorSignal` to write crash stacktraces into the
component's log file (and also to stderr)
- Fix `LOG_MESSAGE` macro: wrap in `do { } while(0)` to prevent
dangling-else
- Fix typo: `file_loggger` → `file_logger`
**Config** (`config.h`, `config.cpp`)
- Add `logging_dir` field to `CliceConfig`, defaulting to
`<cache_dir>/logs/`
- Apply `${workspace}` variable substitution to `logging_dir`
**Master server** (`master_server.h`, `master_server.cpp`)
- After config loads, create a session directory named
`<timestamp>_<pid>` under `logging_dir` and switch master to file
logging
- Pass session log directory to worker pool
**Worker pool** (`worker_pool.h`, `worker_pool.cpp`)
- Pass `--worker-name` (e.g. `SF-0`, `SL-1`) and `--log-dir` to spawned
worker processes
- Add `log_dir` to `WorkerPoolOptions`
**Workers** (`stateful_worker.h/cpp`, `stateless_worker.h/cpp`)
- Accept `worker_name` and `log_dir` parameters; switch to file logging
when `log_dir` is provided
**CLI cleanup** (`clice.cc`)
- Remove `--stateful-worker-count`, `--stateless-worker-count` from CLI
(config-file only)
- Group internal worker args (`--worker-memory-limit`, `--worker-name`,
`--log-dir`) separately
**Docs** (`docs/clice.toml`)
- Fix `logging_dir` example: `.clice/logging` → `.clice/logs`
## Test plan
- [x] `pixi run cmake-build RelWithDebInfo` compiles successfully
- [ ] Verify log files created under `.clice/logs/<timestamp>_<pid>/`
- [ ] Verify each component writes to its own file
- [ ] Verify crash stacktrace appears in component log file
- [ ] Verify `logging_dir` override in `clice.toml` works
- [ ] Verify graceful fallback when log directory is not writable
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Session-specific logging directories (timestamped) and per-worker log
files
* CLI options to set worker name and log directory; general log level
control
* Configurable logging directory with default `<cache_dir>/logs/`
* **Bug Fixes**
* Fixed file-logging name/initialization issues; ensures directory
creation and deterministic filenames
* Added crash-handler support to append stack traces to logs
* **Documentation**
* Updated example config to use `${workspace}/.clice/logs`
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
- **Remove xmake build system**: delete `xmake.lua`, `test-xmake.yml`
workflow, and all xmake-related pixi tasks
- **Migrate packaging to CMake**: `publish-clice.yml` now uses `cmake
-DCLICE_RELEASE=ON` instead of xmake pack
- **Clean up**: remove unused `tests/uv.lock`, `.xmake/` from
`.gitignore`, xmake references from docs
- **Benchmark CI**: change trigger from `pull_request` to
`workflow_dispatch` (manual only)
- **Simplify CLI parsing**: use `DecoKV(style =
KVStyle::JoinedOrSeparate)` in `clice.cc` and `unit_tests.cc`, replacing
verbose `DecoKVStyled` with manual `static_cast` bitmask; use comma
separators; explicit `names` only for underscore fields
## Test plan
- [x] `pixi run cmake-build RelWithDebInfo` compiles successfully
- [x] Verify `pixi run test` passes
- [x] Verify `pixi run package` produces correct archives via CMake
release build
- [x] Verify benchmark workflow can be triggered manually via `gh
workflow run benchmark`
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **Chores**
* Consolidated CI workflows; some automated triggers converted to manual
invocation
* Standardized shell for workflow steps and removed legacy build
workflow
* Switched packaging/build tasks to a CMake/Ninja flow and updated
artifact paths
* Adjusted ignore rules to include previously-ignored build metadata
* **Documentation**
* Removed XMake-specific build and test instructions; docs now reflect
the CMake-based workflow
* **Style**
* Updated CLI option declaration style (no user-facing flag name
changes)
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
PCH and PCM artifacts are now cached to disk at
`.clice/cache/{pch/,pcm/}` with content-addressed filenames, so they
survive server restarts. Dependency metadata is persisted in
`cache.json` (using eventide serde) with a shared path table for
deduplication.
### Key changes
- **protocol.h**: `output_path` field on `BuildPCHParams` /
`BuildPCMParams` so master specifies where workers write
- **stateless_worker.cpp**: Atomic write via `.tmp` + `fs::rename`;
`CompilationUnit` destroyed before rename to flush the file to disk;
fallback to temp file when `output_path` is empty (unit tests)
- **master_server.h**: `PCMState` struct, `pcm_states` map,
`load_cache()` / `save_cache()` / `cleanup_cache()` methods
- **master_server.cpp**: Cache lifecycle — load from `cache.json` on
startup, save after each PCH/PCM build and on exit; deterministic path
computation (`xxh3` preamble hash for PCH, module name + source path
hash for PCM); stale files (>7 days) cleaned on startup; `cache.json`
uses shared path table to avoid redundant storage of header paths across
entries
- **filesystem.h**: `fs::rename()` helper; `ThreadSafeFS` broadened to
match `.pch` extension instead of `preamble-` prefix
- **tests**: 11 new integration tests covering PCH/PCM persistence,
cross-session reuse, staleness detection, shared preamble dedup, and
restart survival; unit tests updated with `output_path`
### Naming scheme
- **PCH**: `.clice/cache/pch/<016x(xxh3(preamble))>.pch`
- **PCM**:
`.clice/cache/pcm/<module_name>-<016x(xxh3(source_path))>.pcm`
## Test plan
- [x] Unit tests — 448 passed
- [x] Integration tests — 92 passed (including 11 new persistent cache
tests)
- [x] Smoke tests — 1 passed
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Fix two data races in the stateful worker that caused spurious
"redefinition" errors during rapid edits, and remove a didChange
workaround that is no longer needed after clice-io/eventide#95.
### stateful_worker.cpp
**Compile handler**: move `params` → `doc` field copy **after**
`strand.lock()`. Previously the copy happened before the lock, so a
concurrent Compile request waiting on the strand could overwrite
`doc.text` while `et::queue` was reading it on the thread pool:
```
T1: Compile A → doc.text = text_A → lock → et::queue reads doc.text
T2: Compile B → doc.text = text_B → waits for strand (overwrites!)
T3: et::queue sees text_B instead of text_A → PCH/text mismatch
```
**DocumentUpdate handler**: only mark `dirty`, stop modifying
`doc.text`/`doc.version`. The event loop notification can fire while
`et::queue` work is running on the thread pool — writing `doc.text` from
one thread while reading it from another is a data race.
### master_server.cpp
Remove the `{0,0}-{0,0}` range workaround for whole-document
`didChange`. eventide's variant deserialization now correctly rejects
`TextDocumentContentChangePartial` when the `range` field is absent
(clice-io/eventide#95), so `TextDocumentContentChangeWholeDocument` is
matched as intended.
### protocol.h
Remove `text` field from `DocumentUpdateParams` — the worker no longer
needs it since DocumentUpdate only sets the dirty flag.
### Integration tests (+312 lines)
Extend test_staleness.py from 5 to 14 tests covering document lifecycle:
- `didChange` body edit → recompilation with updated diagnostics
- `didChange` preamble edit → PCH rebuild + clean recompilation
- `didClose` + reopen → compiles fresh from disk
- `didClose` → hover returns None
- `didSave` header → dependent file recompiles
- `didSave` module → CompileGraph dependents invalidated
## Test plan
- [x] 422 unit tests pass (426 on CI with extra test suites)
- [x] 14 integration tests pass locally
- [x] Depends on clice-io/eventide#95 (merged)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
* **New Features**
* Smaller document-update notifications sent to background workers (only
path and version).
* **Bug Fixes**
* Reduced races and unnecessary work between update and compile flows.
* Prevented notifications from overwriting in-memory document text,
improving state consistency.
* Safer concurrent handling to avoid mid-request eviction of active
documents.
* **Tests**
* Added integration tests for staleness, dependency propagation, and LSP
lifecycle.
* Updated unit tests to match revised update behavior.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Rewrites the template resolver to eliminate
`SubstType`/`CodeSynthesisContexts` dependency, fixing widespread
crashes on real-world C++ code.
### What changed
**Architecture**: replaced double-TreeTransform (PseudoInstantiator +
SubstType) with single-layer design:
- **`SubstituteOnly`** — new lightweight TreeTransform for Phase 2
(typedef expansion + parameter substitution). Does NOT override
`TransformDependentNameType`, breaking the typedef ↔ lookup infinite
cycle.
- **`PseudoInstantiator`** — retains heuristic lookup (the unique value
clang doesn't provide), delegates substitution to `SubstituteOnly`.
**Deleted**:
- `DesugarOnly` class
- `instantiate()` method and all `CodeSynthesisContexts` / `SubstType`
usage
- `state()` / `rewind()` stack management
- `std::abort()` on valid NNS kinds
- `#ifndef NDEBUG` debug flag + `std::print` logging
**Added**:
- `SubstituteOnly` class with depth guard
- `InstantiationStack::findArgument()` — depth/index based parameter
lookup
- CTD→TST resolution for `DependentTemplateSpecializationType` (enables
`__alloc_traits::rebind<T>::other` resolution)
- `active_resolutions` (DNT cycle detection) + `active_ctd_lookups` (CTD
cycle detection via RAII guard)
- Stack frame pollution fix: pop lookup frames before further
`TransformType`
- Pack argument support (single-element forwarding)
- Null safety on all Transform return paths
- Structured `LOG_DEBUG` trace logging with indentation
- `--log-level` / `--test-filter` CLI options for unit test runner
- Bounds checks in `hole()`, `ResugarOnly`, `visitTemplateDeclContexts`
**Tests**: 20 → 36 passing test cases (+5 documented TODOs for known
limitations). New coverage: recursive base classes, multiple
inheritance, typedef chains, CRTP, `remove_reference` partial specs,
`std::map`, `std::basic_string`, pack forwarding.
### Stress test result
```
CDB: llvm-project build (4669 C++ files)
Types resolved: 3,690,190
Types unchanged: 75,907,298
Crashes: 0
```
Before this PR, the same test produced ~52% crash rate (413 crashes in
800 files).
### Known limitations (documented as TODOs)
- NTTP partial specialization matching (`enable_if<true, X>`, `A<X, 0>`)
- Template template parameter deduction
- Non-dependent qualifier nested class templates
(`Outer<int>::Inner<X>`)
- Multi-element pack expansion
- `CXXDependentScopeMemberExpr` lookup (unimplemented)
- Operator-name lookup in dependent contexts
🤖 Generated with [Claude Code](https://claude.com/claude-code)
<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit
## Release Notes
* **New Features**
* Enhanced template resolution with improved cycle detection to prevent
infinite loops
* Better type substitution handling for complex dependent types
* **Bug Fixes**
* Fixed edge cases in template specialization resolution
* Improved null-safety in type transformations
* Enhanced handling of standard library template traits
* **Tests**
* Expanded test coverage for recursive and complex template patterns
* Added validation for standard library type resolution
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Replace the `didSave` sledgehammer (`pch_hashes.clear()` +
mark-all-dirty) with precise per-file dependency tracking that avoids
unnecessary recompilation.
### Two-layer staleness detection
After each successful compilation, a `DepsSnapshot` is captured
(interned path IDs + xxh3 content hashes + timestamp). On the next
feature request, `deps_changed()` checks:
- **Layer 1 (fast):** stat each dep, compare mtime against `build_at`.
If mtime is older → skip. Uses strict `<` so same-second modifications
fall through.
- **Layer 2 (precise):** for files with newer mtime, re-hash content and
compare. Catches touch-without-change (git checkout, backup restore)
without false rebuilds.
Special cases: files unreadable at build time (hash=0) always fall
through to Layer 2; disappeared files are detected via stat failure.
### Consolidated PCH state
Scatter of four maps (`pch_paths`, `pch_bounds`, `pch_hashes`,
`pch_building`) → single `PCHState` struct with `path`, `bound`, `hash`,
`deps`, `building` fields. `DepsSnapshot` and `SymbolInfo` moved out of
`MasterServer` to namespace scope.
### Concurrent compilation dedup
- **`ensure_compiled`:** `DocumentState::compiling` event prevents
duplicate AST compilations. Waiters `co_await` the event and check
`ast_dirty` after waking. When deps change is detected during an
in-flight build, `generation` is bumped so the builder's generation
check prevents it from incorrectly clearing `ast_dirty`.
- **`ensure_pch`:** `PCHState::building` event deduplicates PCH builds.
Waiters re-validate `preamble_hash` after waking to handle edits during
the wait. The `bound==0` path waits for in-flight builds before erasing.
Old PCH is deleted and path cleared before rebuild starts, so waiters
never see a stale path on failure.
### `didSave` changes
Removed the blanket `pch_hashes.clear()` + mark-all-dirty on save.
Staleness is now detected lazily via `deps_changed()` at the next
feature request. `didSave` still invalidates `CompileGraph` dependents
for module deps.
### FIXME noted
Rapid `didChange` edits (especially preamble changes) can cause the
stateful worker to compile with stale/concatenated text. Root cause is
in the worker, not in staleness tracking — noted as FIXME for a
follow-up PR.
## Test plan
13 integration tests covering:
- [x] Header mtime change → AST recompilation
- [x] Preamble header change → PCH rebuild
- [x] No change → fast path (cached AST reused)
- [x] Touch without content change → Layer 2 hash skips recompile
- [x] Header replaced with different content → detected
- [x] Fix error in header → diagnostics clear
- [x] Multiple files sharing header → each detects independently
- [x] Transitive header change → detected through include chain
- [x] didChange body edit → recompilation with updated diagnostics
- [x] didClose + reopen → compiles new disk content
- [x] didClose → hover returns None
- [x] didSave header → dependent file recompiles
- [x] didSave module → CompileGraph dependents invalidated
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## Summary
Replace the push-based compilation model with a pull-based (lazy) model
where compilation is driven entirely by feature requests.
### Server core (`master_server.cpp/h`)
- **Remove** `schedule_build()`, `run_build_drain()`, debounce timers,
and `DocumentState` flags (`build_running`, `build_requested`,
`drain_scheduled`)
- **Remove** `debounce_ms` config field
- `didOpen`/`didChange` only update `DocumentState` and mark `ast_dirty`
— no compilation triggered
- `didSave` marks dependent docs dirty via `CompileGraph::update()`,
invalidates PCH hashes, marks **all** open documents `ast_dirty` (header
saves), and queues background indexing
- **Implement** `ensure_compiled(path_id)` — the pull-based entry point
called by `forward_stateful()`/`forward_stateless()` before every
feature request:
1. Fast-path if `!ast_dirty`
2. Compile C++20 module deps via `compile_graph->compile_deps()`
3. Build/reuse PCH via `ensure_pch()` (only attach on success)
4. Send `CompileParams` to stateful worker
5. Publish diagnostics, clear dirty, schedule indexing
6. Generation mismatch → return `false`, keep dirty for retry
- `forward_stateless()` now also calls `compile_graph->compile_deps()`
before stateless requests (completion/signatureHelp)
- Move module-implementation-unit implicit dependency handling into
`resolve_fn` (was duplicated in `run_build_drain` and `ensure_compiled`)
### CompileGraph (`compile_graph.cpp/h`)
- **Add** `compile_deps(path_id)` — compiles all transitive module
dependencies but NOT the file itself (used for plain .cpp files that
`import` modules)
- Unify `compile`/`compile_deps` via `compile_impl(path_id, ancestors,
dispatch_self)` parameter
- `compile_deps` compiles dependencies concurrently via `when_all`
- Extract `finish()` lambda to deduplicate `compiling=false;
completion->set()` cleanup across all exit paths
- Use `std::ranges::remove` instead of legacy `std::remove`
### Test infrastructure (`conftest.py`)
- `open_and_wait()` now sends a hover request to trigger
`ensure_compiled()` (pull-based model requires a feature request to
compile)
- Fix URI handling: send percent-encoded URI on the wire, normalize for
internal lookups, store diagnostics under both raw and normalized URI
keys
- Add `_normalize_uri()` helper using `urllib.parse.unquote`
### Integration tests
- Update all tests for pull-based model: no more waiting on `didOpen`
diagnostics
- `_wait_for_index()` sends hover to trigger compilation before polling
`workspace/symbol`
- `test_hover_save_close` simplified — hover directly triggers
compilation
- `test_save_recompile` and `test_pch_*` wait for fresh diagnostics
after hover-triggered recompilation
### Unit tests (`compile_graph_tests.cpp`)
- Extract `compiled`/`graph` as TEST_SUITE members with
`std::optional<CompileGraph>`
- Extract `execute(callback)` helper to deduplicate event_loop
boilerplate
- Add 8 new `compile_deps` tests: no-deps, single dep, chain, diamond,
failure, plain-cpp, concurrent dedup, resolve-once
- Remove redundant `inline` on file-scope helpers
## Test plan
- [x] Unit tests: 426 passed, 5 skipped
- [x] Smoke tests: 1/1 passed
- [x] Integration tests: 69 passed, 0 failed, no hangs
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
## What changed
This PR moves the shared VS Code development workflow to the repository
root.
It adds repo-level `.vscode/launch.json` and `.vscode/tasks.json` so
contributors can debug `clice` and the VS Code extension from the root
workspace with relative paths.
The checked-in tasks are intentionally limited to extension-side `pnpm`
workflows such as installing dependencies and running the extension
watcher. They do not build the `clice` executable or unit tests.
It also updates `.gitignore` to allow only those two shared VS Code
config files to be committed, while continuing to ignore other local
`.vscode` files.
Finally, it removes the duplicate `editors/vscode/.vscode` folder so
there is a single supported VS Code setup in the repo.
## Why
The previous setup split VS Code configs between local-only root files
and tracked configs under `editors/vscode/.vscode`, which made the
development/debug flow less consistent.
Keeping only extension-side tasks avoids baking a specific native build
workflow into committed editor config while still making extension
development usable out of the box.
## Developer impact
Contributors can now open the repository root in VS Code and use the
checked-in launch/tasks configs directly.
The extension debug configs only disable competing C/C++ language
extensions (`clangd` and `cpptools`) instead of disabling every
extension.
All committed paths remain relative to `${workspaceFolder}`.
## Validation
- Ran `python3 -m json.tool .vscode/launch.json`
- Ran `python3 -m json.tool .vscode/tasks.json`
