alex/g3
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update dependency analysis artifacts with detailed evidence

Browse Source 
- hotspots.md: Added specific dependent file lists for each hotspot
- hotspots.md: Added cross-crate coupling points table
- hotspots.md: Added crate-level coupling scores
- limitations.md: Expanded coverage of unobservable patterns
- limitations.md: Added confidence levels for inferences
- limitations.md: Added extraction method details table

Agent: euler
This commit is contained in:
Dhanji R. Prasanna
2026-02-02 17:20:15 +11:00
parent 263a838d31
commit 95d9847354
2 changed files with 153 additions and 96 deletions
Download Patch File Download Diff File Expand all files Collapse all files

114
analysis/deps/hotspots.md
View File

@@ -1,69 +1,79 @@
# Coupling Hotspots
**Generated**: 2025-02-02
**Method**: Fan-in/fan-out analysis from dependency graph
## High Fan-In Files (Most Depended Upon)
Files with disproportionately high incoming dependencies.
These files are imported by many other files. Changes here have wide impact.
| File | Fan-In | Crate | Role |
|------|--------|-------|------|
| ui_writer.rs | 15 | g3-core | UI abstraction trait |
| g3_status.rs | 10 | g3-cli | Status message formatting |
| simple_output.rs | 8 | g3-cli | Output helper |
| context_window.rs | 6 | g3-core | Token/context management |
| template.rs | 6 | g3-cli | Template processing |
| paths.rs | 5 | g3-core | Path utilities |
### Evidence: ui_writer.rs (fan-in: 15)
Depended on by:
- g3-cli: accumulative.rs, agent_mode.rs, autonomous.rs, commands.rs, interactive.rs, task_execution.rs, ui_writer_impl.rs, utils.rs
- g3-core: compaction.rs, feedback_extraction.rs, lib.rs, retry.rs, tool_dispatch.rs, tools/*.rs
### Evidence: g3_status.rs (fan-in: 10)
Depended on by:
- commands.rs, interactive.rs, simple_output.rs, task_execution.rs, and others in g3-cli
| File | Fan-In | Dependents |
|------|--------|------------|
| `g3-core/src/lib.rs` | 18 | streaming_parser, context_window, acd, streaming, stats, retry, feedback_extraction, task_result, tool_dispatch, tools/* |
| `g3-core/src/ui_writer.rs` | 14 | retry, compaction, feedback_extraction, tool_dispatch, tools/{acd,executor,shell,research,file_ops,plan,memory,misc,webdriver} |
| `g3-cli/src/simple_output.rs` | 9 | utils, interactive, autonomous, coach_feedback, accumulative, task_execution, commands, agent_mode |
| `g3-cli/src/template.rs` | 6 | project_files, accumulative, commands, embedded_agents, agent_mode, interactive |
| `g3-core/src/paths.rs` | 6 | acd, session, context_window, tools/{executor,shell,plan} |
| `g3-core/src/context_window.rs` | 5 | compaction, stats, streaming, session, task_result |
| `g3-cli/src/g3_status.rs` | 5 | simple_output, interactive, task_execution, commands |
| `g3-cli/src/display.rs` | 4 | interactive, ui_writer_impl, agent_mode |
| `g3-cli/src/ui_writer_impl.rs` | 4 | autonomous, coach_feedback, accumulative, agent_mode |
| `g3-core/src/utils.rs` | 3 | tools/{shell,file_ops} |
## High Fan-Out Files (Most Dependencies)
Files with disproportionately high outgoing dependencies.
These files import many other modules. They are integration points.
| File | Fan-Out | Crate | Role |
|------|---------|-------|------|
| agent_mode.rs | 13 | g3-cli | Agent mode entry point |
| lib.rs | 13 | g3-core | Core library root |
| commands.rs | 12 | g3-cli | Command handlers |
| interactive.rs | 12 | g3-cli | Interactive REPL |
| accumulative.rs | 11 | g3-cli | Accumulative mode |
| planner.rs | 8 | g3-planner | Planning orchestration |
| File | Fan-Out | Dependencies |
|------|---------|-------------|
| `g3-cli/src/interactive.rs` | 11 | g3-core, completion, commands, display, g3_status, project, project_files, simple_output, input_formatter, template, task_execution, utils |
| `g3-cli/src/agent_mode.rs` | 11 | g3-core, project_files, display, language_prompts, simple_output, embedded_agents, ui_writer_impl, interactive, template, project, cli_args |
| `g3-cli/src/accumulative.rs` | 8 | g3-core, autonomous, cli_args, interactive, simple_output, ui_writer_impl, utils, template |
| `g3-cli/src/commands.rs` | 7 | g3-core, completion, g3_status, simple_output, project, template, task_execution |
| `g3-core/src/tools/executor.rs` | 7 | g3-config, background_process, pending_research, paths, ui_writer, webdriver_session, lib (ToolCall) |
| `g3-cli/src/autonomous.rs` | 5 | g3-core, coach_feedback, metrics, simple_output, ui_writer_impl |
| `g3-core/src/compaction.rs` | 4 | g3-providers, context_window, provider_config, ui_writer |
| `g3-core/src/tool_dispatch.rs` | 4 | tools/executor, tools/mod, ui_writer, lib (ToolCall) |
| `g3-cli/src/ui_writer_impl.rs` | 4 | g3-core, filter_json, display, streaming_markdown |
| `g3-core/src/streaming.rs` | 4 | g3-providers, context_window, streaming_parser, lib (ToolCall) |
### Evidence: agent_mode.rs (fan-out: 13)
## Cross-Crate Coupling Points
Depends on:
- g3-core: Agent, ui_writer::UiWriter
- Internal: project_files, display, language_prompts, simple_output, embedded_agents, ui_writer_impl, interactive, template
Files that bridge multiple crates:
### Evidence: g3-core/lib.rs (fan-out: 13)
| File | Cross-Crate Imports | Crates Touched |
|------|---------------------|----------------|
| `g3-core/src/lib.rs` | 2 | g3-config, g3-providers |
| `g3-core/src/webdriver_session.rs` | 1 | g3-computer-control |
| `g3-core/src/tools/webdriver.rs` | 1 | g3-computer-control |
| `g3-core/src/tools/executor.rs` | 1 | g3-config |
| `g3-core/src/tools/research.rs` | 1 | g3-config |
| `g3-core/src/provider_registration.rs` | 2 | g3-config, g3-providers |
| `g3-planner/src/llm.rs` | 3 | g3-config, g3-core, g3-providers |
Depends on:
- External crates: g3-config, g3-providers
- Internal modules: ui_writer, context_window, paths, compaction, streaming, tools, etc.
## Crate-Level Coupling
## Cross-Crate Coupling
| Source Crate | Target Crate | Edge Count |
|--------------|--------------|------------|
| g3-cli | g3-core | 35 |
| g3-core | g3-providers | 10 |
| g3-core | g3-config | 5 |
| g3-planner | g3-core | 4 |
| g3-planner | g3-providers | 3 |
| g3-core | g3-computer-control | 2 |
| Crate | Outgoing Deps | Incoming Deps | Coupling Score |
|-------|---------------|---------------|----------------|
| g3-cli | 5 | 1 | High (consumer) |
| g3-core | 4 | 3 | High (hub) |
| g3-planner | 3 | 1 | Medium |
| g3-providers | 0 | 5 | High (foundation) |
| g3-config | 0 | 5 | High (foundation) |
| g3-computer-control | 0 | 2 | Low |
| g3-execution | 0 | 1 | Low |
| studio | 0 | 0 | Isolated |
## Observations
1. **ui_writer.rs** is a central abstraction point; changes here affect 15+ files
2. **g3-cli** files have high fan-out due to orchestration responsibilities
3. **g3-core/lib.rs** is the primary API surface with expected high coupling
4. **g3_status.rs** and **simple_output.rs** form a UI utility cluster in g3-cli
5. **tools/*.rs** files consistently depend on ui_writer and ToolCall types
1. **g3-core/src/lib.rs** is the primary coupling hotspot. It defines `Agent`, `ToolCall`, and other core types used throughout the codebase.
2. **g3-core/src/ui_writer.rs** defines the `UiWriter` trait used by all tool implementations for output. High fan-in is expected for a trait definition.
3. **g3-cli/src/simple_output.rs** is a utility wrapper used by most CLI modules. High fan-in indicates it's a well-factored common dependency.
4. **g3-cli/src/interactive.rs** and **g3-cli/src/agent_mode.rs** have the highest fan-out, as expected for top-level orchestration modules.
5. **g3-providers** and **g3-config** are foundation crates with zero outgoing dependencies and high incoming dependencies. This is the expected pattern for leaf crates.
6. **studio** is completely isolated from the g3 crate ecosystem.

135
analysis/deps/limitations.md
View File

@@ -1,67 +1,114 @@
# Analysis Limitations
## What Was Observed
| Category | Method | Coverage |
|----------|--------|----------|
| Crate dependencies | Cargo.toml parsing | Complete |
| Module structure | `mod` declarations in lib.rs/main.rs | Complete |
| File imports | `use g3_*` and `use crate::` patterns | Partial |
| Submodule structure | mod.rs file inspection | Complete |
**Generated**: 2025-02-02
## What Could Not Be Observed
### 1. Implicit Dependencies
### 1. Dynamic/Runtime Dependencies
- **Trait implementations**: A file implementing a trait from another module creates a dependency not captured by `use` statements
- **Type aliases**: Types re-exported through `pub use` chains
- **Derive macros**: Dependencies introduced by `#[derive(...)]`
- **Plugin loading**: Any runtime module loading via `dlopen` or similar
- **Reflection-based imports**: Dependencies resolved at runtime
- **Configuration-driven imports**: Modules loaded based on config values
### 2. Dynamic Patterns
### 2. Macro-Generated Code
- **Trait objects**: `dyn Trait` usage creates runtime dependencies
- **Generic bounds**: `T: SomeTrait` constraints
- **Associated types**: Dependencies through associated type resolution
- **Procedural macros**: Code generated by `#[derive(...)]` and custom proc macros
- **Declarative macros**: `macro_rules!` expansions that generate `use` statements
- **Build script outputs**: Code generated by `build.rs` files
### 3. Conditional Compilation
- **Platform-specific code**: `#[cfg(target_os = "...")]` blocks not evaluated
- **Feature flags**: `#[cfg(feature = "...")]` dependencies not traced
- **Test-only code**: `#[cfg(test)]` modules excluded
- **Platform-specific code**: `#[cfg(target_os = "...")]` blocks are included regardless of target
- **Feature flags**: `#[cfg(feature = "...")]` gated code is included unconditionally
- **Test-only code**: `#[cfg(test)]` modules are excluded from analysis
### 4. Build System
### 4. Re-exports and Transitive Dependencies
- **build.rs**: Build script dependencies not analyzed
- **Procedural macros**: Macro crate dependencies not traced
- **Workspace-level features**: Feature unification effects not modeled
- **`pub use` chains**: Transitive re-exports are not fully traced
- **Glob imports**: `use module::*` does not enumerate specific items
- **Prelude imports**: Implicit prelude items are not tracked
### 5. Excluded Files
### 5. External Crate Dependencies
| Category | Count | Reason |
|----------|-------|--------|
| Test files (`/tests/`) | ~40 | Out of scope |
| Example files (`/examples/`) | ~10 | Out of scope |
| Worktree copies | ~100 | Duplicates |
- **Third-party crates**: Only internal g3 workspace crates are analyzed
- **Workspace dependency versions**: Version constraints not analyzed
- **Optional dependencies**: Not distinguished from required dependencies
## What Was Inferred
| Inference | Basis | Confidence |
|-----------|-------|------------|
| Layer assignment | Topological sort of crate deps | High |
| Fan-in/fan-out metrics | Edge counting | High |
| Module-to-file mapping | Naming convention (mod.rs, *.rs) | High |
| Entrypoints | lib.rs/main.rs presence | High |
### 1. Module Hierarchy
Module parent-child relationships inferred from:
- `mod` declarations in `lib.rs` and `mod.rs` files
- File system structure (`foo/mod.rs` implies `foo` module)
**Confidence**: High - Rust module system is deterministic
### 2. Crate Dependencies
Crate-level dependencies extracted from:
- `Cargo.toml` `[dependencies]` sections
- `path = "..."` declarations for workspace crates
**Confidence**: High - Cargo.toml is authoritative
### 3. File-Level Imports
File imports extracted via regex pattern matching:
- `^use (g3_|crate::)` for internal imports
- `^(pub )?mod ` for module declarations
**Confidence**: Medium-High - May miss unusual import patterns
### 4. Fan-In/Fan-Out Counts
Coupling metrics derived from edge counts in dependency graph.
**Confidence**: Medium - Depends on completeness of edge extraction
## What May Invalidate Conclusions
1. **Hidden re-exports**: If `g3-core/lib.rs` re-exports types from submodules, actual coupling may be higher
2. **Macro-generated code**: Macros like `#[derive(Serialize)]` add implicit serde dependency
3. **Runtime plugin loading**: If any crate loads code dynamically, static analysis misses it
4. **Workspace member changes**: Adding/removing crates from workspace invalidates crate-level graph
### 1. Code Changes
## Recommendations for Improved Analysis
This analysis reflects the codebase state at generation time. Any of the following invalidate the analysis:
- New files added
- Import statements changed
- Module structure reorganized
- Cargo.toml dependencies modified
1. Use `cargo metadata` for authoritative crate graph
2. Use `cargo tree` for transitive dependency analysis
3. Use `rust-analyzer` for semantic import resolution
4. Parse AST for type references beyond `use` statements
### 2. Incomplete Extraction
The regex-based extraction may miss:
- Multi-line `use` statements with unusual formatting
- Imports inside function bodies
- Imports in macro invocations
- `extern crate` declarations (deprecated but valid)
### 3. Test Code Exclusion
Test files (`tests/*.rs`, `#[cfg(test)]` modules) are excluded. Test-only dependencies and coupling patterns are not reflected.
### 4. Example Code Exclusion
Example files (`examples/*.rs`) are excluded from the graph.
### 5. Build Script Dependencies
`build.rs` files may introduce compile-time dependencies not reflected in the import graph.
## Extraction Method Details
| Aspect | Method | Tool |
|--------|--------|------|
| Crate dependencies | Cargo.toml parsing | Manual inspection |
| Module declarations | Regex: `^(pub )?mod ` | ripgrep |
| Internal imports | Regex: `^use (g3_\|crate::)` | ripgrep |
| File enumeration | `find crates -name "*.rs"` | find |
| Cycle detection | Manual graph traversal | N/A |
## Recommendations for Future Analysis
1. **Use cargo metadata**: `cargo metadata --format-version 1` provides authoritative dependency information
2. **Use rust-analyzer**: LSP-based analysis would capture macro expansions
3. **Use cargo-depgraph**: Specialized tool for Rust dependency visualization
4. **Include test coverage**: Analyze test files separately for test-specific coupling