- is_typing_declaration_name(): had the same Token::Ident bug as
determine_context() — matched string keywords against Token::Ident
but the lexer produces dedicated variants. Extracted shared
is_declaration_keyword() helper that matches actual token variants.
- determine_context(): add Token::Schedule so schedule blocks get
InFieldBlock context instead of falling through to Unknown.
- Tests: add negative assertions to test_top_level_completions,
test_field_block_completions, and test_relationship_completions.
Add test_no_completions_after_declaration_keyword,
test_no_completions_while_typing_name, and
test_schedule_completions_no_toplevel.
The Logos lexer produces dedicated token variants for all keywords
(Token::Behavior, Token::LifeArc, etc.) — Token::Ident is only for
user-defined identifiers. determine_context() was matching against
Token::Ident(keyword), so last_keyword was always None and every
nested block returned CompletionContext::Unknown, causing
all_keyword_completions() to dump top-level declaration keywords
everywhere.
Changes:
- determine_context(): match actual token variants (Token::Behavior,
Token::LifeArc, Token::Relationship, Token::Character, etc.) and
track last_context directly instead of a String keyword
- behavior_keyword_completions(): add all word-based BT keywords
(choose, then, repeat, if, when, invert, retry, timeout, cooldown,
succeed_always, fail_always)
- InFieldBlock arm: always include field_keyword_completions() so
generic fields (age, bond, etc.) appear even without a species
- Tests: add negative assertions confirming top-level keywords are
absent inside behavior/life_arc blocks; add new test
test_behavior_no_toplevel_keywords for nested block isolation
Zed extensions do not automatically use queries from fetched grammars.
They require local query files in languages/<lang>/ directory.
Restored highlights.scm by copying from tree-sitter-storybook/queries/.
The regression test ensures this stays in sync with the source of truth.
Also updated grammar revision to 0590c13 (latest remote commit).
Added empty [workspace] table to zed-storybook/Cargo.toml to explicitly
tell cargo this package is not part of the parent workspace. This fixes
the Zed extension build error.
Added zed-storybook to workspace.exclude to prevent cargo from
treating it as a workspace member. This allows Zed to build the
extension independently.
This commit completes the migration started in the previous commit,
updating all remaining files:
- Lexer: Changed token from Extends to Modifies
- Parser: Updated lalrpop grammar rules and AST field names
- AST: Renamed Schedule.extends field to modifies
- Grammar: Updated tree-sitter grammar.js
- Tree-sitter: Regenerated parser.c and node-types.json
- Examples: Updated baker-family work schedules
- Tests: Updated schedule composition tests and corpus
- Docs: Updated all reference documentation and tutorials
- Validation: Updated error messages and validation logic
- Package: Bumped version to 0.3.1 in all package manifests
All 554 tests pass.
Changed the schedule composition keyword from "extends" to "modifies"
to better reflect the semantic meaning of schedule inheritance. When
a schedule modifies another, it inherits base blocks and can override
them by name or add new blocks.
This is a breaking change for all existing Storybook files that use
schedule composition. The migration is a simple find-and-replace:
schedule X extends Y → schedule X modifies Y
Changes include:
- Grammar: Updated tree-sitter grammar and lexer token
- Parser: Updated lalrpop parser and AST field names
- Documentation: Updated all reference docs, tutorials, and specs
- Examples: Updated baker-family example schedules
- Tests: Updated all test cases and corpus files
- Testing: Added type system keywords to prop_tests exclusion list
- Tooling: Added xtask for workspace cleanup
- Version: Bumped to v0.3.1 (skipping v0.3.0)
- Spec: Created SBIR v0.3.1 spec documenting the change
BREAKING CHANGE: The "extends" keyword for schedules has been
replaced with "modifies". Update all schedule declarations.
Updated tree-sitter grammar to match v0.3.0 LALRPOP parser:
Grammar updates:
- Schedule: block-based syntax with extends, override, recurrence
- Life arc: requires clause for field validation
- Template: uses behaviors/schedules syntax
- Behavior: correct keywords (choose/then/repeat with optional params)
- Type system: concept_comparison with any/is_condition
- Removed concept semicolon requirement
Query file updates:
- highlights.scm: updated node names to *_declaration
- outline.scm: updated for new declaration node names
- indents.scm: updated node names, removed concept semicolon
Corpus test updates:
- Created schedules.txt with v0.3.0 syntax tests
- Created highlights.txt for highlighting tests
- Updated type_system.txt for v0.3.0 type syntax
- Updated behaviors.txt for correct expression wrapping
- Updated declarations.txt to use correct node names
- Updated basic.txt to use character_declaration/character_body
- Deleted obsolete v0.2.0 syntax tests
Integration tests:
- Added tree_sitter_integration.rs test suite
- Fixed test_any_type to use correct v0.3.0 syntax
- Fixed test_tree_sitter_grammar_builds to use generate command
Changed highlights.scm to use (any_type) node type instead of the
string literal "any". The grammar defines any as any_type rule,
not as a bare keyword string.
The tests were using line: 2 but the character declarations were on
line: 1 (due to the leading newline in the raw string literal). This
caused the cursor position to be outside the character span, making
the code actions fail to trigger.
Fixed by changing line: 2 to line: 1 in both test_convert_species_to_template
and test_convert_template_to_species.
Added "What's New in v0.3" section covering species, concepts,
sub_concepts, concept_comparison, template species inheritance,
and life arc field requirements. Updated quick start example
with v0.3 syntax including species and type system declarations.
Added tests verifying semantic token highlighting for concept,
sub_concept (enum and record forms), and concept_comparison
declarations. Made token_type_index pub(crate) for test access.
Updated binary format spec for v0.3.0 type system changes:
- Section 3 (Types): concept, sub_concept, concept_comparison encoding
- Section 4 (Characters): Value/Expression discriminant renames
- Section 5 (Templates): species_base field for inheritance
- Section 12 (Life Arcs): required_fields with type annotations
- Section 13 (Enums): note on sub_concept enum distinction
- Changelog and version history updated
Also fixed clippy/import issues in LSP semantic tokens module.
Added top-level keyword completions for:
- concept: with semicolon-terminated snippet
- sub_concept: with dot notation snippet
- concept_comparison: with brace-delimited snippet
Added test verifying type system keywords appear in
completions with correct snippet formatting.
Updated extension.toml:
- Version bumped to 0.3.0
- Grammar rev updated to v0.3 tree-sitter commit
Updated Zed highlights.scm with type system support:
- concept, sub_concept, concept_comparison keywords
- any keyword highlighting
- Named field highlights for type system declarations
- Template species reference highlighting
- sub_concept_field property highlighting
Added optional requires clause to life_arc declarations:
life_arc Baker requires { skill_level: Number } { ... }
Includes new FieldRequirement AST type, requires keyword token,
and parser support for the requires clause.
Added grammar rules for v0.3 type system declarations:
- concept_declaration: semicolon-terminated base type
- sub_concept: enum and record forms with dot notation
- concept_comparison: nested pattern matching with variant_pattern
- any_type: named node for the 'any' keyword
- template species extension: optional ': Species' syntax
Updated query files:
- highlights.scm: new keywords and named field highlights
- outline.scm: concept/sub_concept/concept_comparison in outline
- indents.scm: indentation for new brace-delimited nodes
Fixed pre-existing query issues:
- Replaced invalid repeat_node reference with decorator_node
- Removed invalid '?' punctuation reference
Added comprehensive test corpus (type_system.txt) covering:
- concept declaration (simple and multiple)
- sub_concept enum and record forms
- concept_comparison with any/is conditions
- template with species extension
- full combined example
Checks that concept_comparison variant names exist in the referenced
concept's enum sub-concepts and that all variants are covered
(exhaustiveness checking).
ConceptRegistry tracks all concepts and sub_concepts, providing lookup
and validation for sub-concept variant/field references. Supports both
enum and record sub-concept kinds.
When a template extends a species via species_base, shared field names
must have compatible types. Range values are compatible with their
underlying numeric types (Number or Decimal).
When a template has a species_base, its fields are loaded first as the
lowest priority layer. Override order: species -> includes -> template
-> character (last-one-wins for field values).
Added optional species_base field to Template struct enabling
template-species inheritance syntax: `template Name: Species { ... }`.
Updated LALRPOP grammar and all Template construction sites.
Added validation that sub_concept declarations reference an existing
parent concept. Produces clear error message with guidance to add
the missing concept declaration.
Changed sub_concept syntax from heuristic parent extraction to
explicit dot notation: `sub_concept Parent.Name { ... }`
The old syntax inferred the parent concept from the name using
uppercase letter heuristics. The new syntax makes the parent
concept explicit and unambiguous.
Added type_system_tests module with comprehensive tests.
Added complete support for the new type system syntax including:
- concept: Base type declarations
- sub_concept: Enum and record sub-type definitions
- concept_comparison: Compile-time pattern matching with conditional guards
Parser changes:
- Added VariantPattern, FieldCondition, and Condition AST nodes
- Implemented "is" keyword for pattern matching (e.g., "CupType is Glass or CupType is Plastic")
- Added Value::Any variant to support universal type matching
- Disambiguated enum-like vs record-like sub_concept syntax
LSP updates:
- Added Value::Any match arms across code_actions, completion, hover, inlay_hints, and semantic_tokens
- Type inference and formatting support for Any values
Example fixes:
- Fixed syntax error in baker-family behaviors (missing closing brace in nested if)
- Removed deprecated core_enums.sb file
Added AST structures for new type system:
- ConceptDecl: Base type definition
- SubConceptDecl: Enum or record subtype with SubConceptKind
- ConceptComparisonDecl: Compile-time enum mapping
- ConceptMapping: Individual mapping entry
Added Declaration enum variants and stubs throughout codebase
for LSP, resolve, and semantic analysis modules.
Added four keywords for new type system:
- concept: Base type definition
- sub_concept: Enum/record sub-type definition
- concept_comparison: Compile-time enum mapping
- any: Universal type for dynamic contexts
Also added:
- CLAUDE.md with project instructions and commit guidelines
- Test coverage for new keywords
- Crate-level deny directives for unused variables and dead code
Fixed pre-existing clippy issues to pass pre-commit hooks.
BREAKING CHANGES:
- Relationship syntax now requires blocks for all participants
- Removed self/other perspective blocks from relationships
- Replaced 'guard' keyword with 'if' for behavior tree decorators
Language Features:
- Add tree-sitter grammar with improved if/condition disambiguation
- Add comprehensive tutorial and reference documentation
- Add SBIR v0.2.0 binary format specification
- Add resource linking system for behaviors and schedules
- Add year-long schedule patterns (day, season, recurrence)
- Add behavior tree enhancements (named nodes, decorators)
Documentation:
- Complete tutorial series (9 chapters) with baker family examples
- Complete reference documentation for all language features
- SBIR v0.2.0 specification with binary format details
- Added locations and institutions documentation
Examples:
- Convert all examples to baker family scenario
- Add comprehensive working examples
Tooling:
- Zed extension with LSP integration
- Tree-sitter grammar for syntax highlighting
- Build scripts and development tools
Version Updates:
- Main package: 0.1.0 → 0.2.0
- Tree-sitter grammar: 0.1.0 → 0.2.0
- Zed extension: 0.1.0 → 0.2.0
- Storybook editor: 0.1.0 → 0.2.0
Separate if(expr) condition from if(expr) { child } decorator to fix
grammar conflicts. Simplify decorator_params to accept only integer,
range, or duration.
Add comprehensive example demonstrating all major features of the
Storybook DSL through Lewis Carroll's Alice in Wonderland.
Contents:
- 12 characters (Alice, WhiteRabbit, CheshireCat, MadHatter, etc.)
- 7 relationships with named participant blocks
- 3 institutions (tea parties, royal court)
- 1 schedule (Mad Tea Party eternal rotation)
- 12 behavior trees using new syntax features
- 1 life arc (Alice's journey)
Demonstrates:
- Cross-file template resolution (schema/templates.sb)
- Template inheritance and includes
- Strict template validation
- Action parameters and repeater decorators
- Named participant relationship blocks
- Species-based character typing
- Rich prose blocks throughout
All content validates successfully.
Add syntax enhancements for more expressive behavior trees and relationships.
Action parameters:
- Support typed/positional parameters: WaitDuration(2.0)
- Support named parameters: SetValue(field: value)
- Enable inline values without field names
Repeater decorator:
- Add * { node } syntax for repeating behavior nodes
- Maps to Decorator("repeat", node)
Named participant blocks:
- Replace self/other blocks with named participant syntax
- Support multi-party relationships
- Example: Alice { role: seeker } instead of self { role: seeker }
Schedule block syntax:
- Require braces for schedule blocks to support narrative fields
- Update tests to use new syntax: 04:00 -> 06:00: Activity { }
Enable characters to inherit from templates defined in different files
across the project structure.
- Add file_index field to NameEntry to track declaration source files
- Update NameTable::from_files() to set file indices when merging tables
- Change conversion pipeline to pass &[ast::File] instead of flat arrays
- Update merge functions to use two-level indexing:
all_files[entry.file_index].declarations[entry.decl_index]
- Update all affected tests to use new signatures
