Standardize private/public method naming conventions

dev/philosophy/modules.md

@@ -92,9 +92,31 @@ end
 Why: visible origin, safe refactors (only the `using ..X` line changes), no accidental
 shadowing, same rule for external and sibling symbols.
 
+## Naming conventions
+
+Three visibility levels, signalled by the name itself:
+
+- `symbol` — **public**: exported via `export`; part of the documented API.
+- `_symbol` — **private helper**: unexported; internal implementation detail.
+  Accessible via qualified path (`Module._symbol`) but not advertised.
+- `__symbol` — **default value**: unexported; provides a replaceable semantic
+  default (e.g. `__display() = true`). The double underscore distinguishes it
+  from a plain helper: it is a *default* that a higher-level layer may override,
+  not merely a utility function.
+
+```julia
+export build_thing           # public
+
+function _validate(x)        # private helper
+    ...
+end
+
+__verbose()::Bool = false    # default value (replaceable by extension or subtype)
+```
+
 ## Two-level exports
 
-- **Submodule**: `export` for its public API; internals (`_helper`) unexported.
+- **Submodule**: `export` for its public API; internals (`_helper`, `__default`) unexported.
 - **Package (top-level)**: **no** `export`. Load submodules with `using .Submodule`;
   users reach them via `Package.Submodule.sym`.
 

dev/planning.md

@@ -6,6 +6,7 @@ to align with the human before touching code, not to produce a 50-step document.
 ## When to write a plan
 
 Before any task that:
+
 - touches more than one file,
 - changes a public interface, or
 - involves a non-obvious design decision.
@@ -16,7 +17,7 @@ Write the plan, share it, wait for confirmation before starting.
 
 ## Template
 
-```
+````markdown
 # Plan: <short title>
 
 ## What and why
@@ -34,32 +35,74 @@ Write the plan, share it, wait for confirmation before starting.
 Steps:
 1. …
 2. …
-Checkpoint: run `test/suite/<group>` via MCP. All green before moving on.
+Checkpoint:
+
+```bash
+# MCP (preferred)
+get_test_command test_args=["suite/<group>"]
+then generate_report
+
+# Fallback
+julia --project=@. -e 'using Pkg; Pkg.test(; test_args=["suite/<group>"])' 2>&1 | tee /tmp/<branch>_phaseA.log
+grep -E "Error|Fail|Test Summary" /tmp/<branch>_phaseA.log
+```
+
+All green before moving on.
 
 ### Phase B — <name>
+
 Steps:
 1. …
-Checkpoint: run `test/suite/<group>` + full suite.
+Checkpoint:
+
+```bash
+# MCP (preferred)
+get_test_command test_args=["suite/<group1>", "suite/<group2>"]
+then generate_report
+
+# Final phase only: full suite
+julia --project=@. -e 'using Pkg; Pkg.test()' 2>&1 | tee /tmp/<branch>_final.log
+grep -E "Error|Fail|Test Summary" /tmp/<branch>_final.log
+```
 
 …
 
 ## Human checkpoints
+
 - ⛔ Ask before first commit.
 - ⛔ Ask before pushing to a shared branch.
 - ⛔ Ask before deleting files that were not created in this task.
 - ⛔ Ask if a design decision arises that was not in the plan.
 
 ## Out of scope
+
 <List things explicitly not done in this task.>
-```
+````
 
 ---
 
 ## Rules for filling in the template
 
 **Phases** — one phase = one coherent unit that can be tested independently. A phase
-ends with a test checkpoint (run the affected test suite via the MCP `get_test_command`
-tool, then `generate_report`). Never skip the checkpoint.
+ends with a test checkpoint. Never skip the checkpoint.
+
+**Test checkpoints** — always use the MCP `get_test_command` tool with `test_args` to
+run only the affected test groups, then `generate_report` to parse the result. If MCP
+is unavailable, fall back to:
+
+```bash
+# Phase-specific (preferred during intermediate phases)
+julia --project=@. -e 'using Pkg; Pkg.test(; test_args=["suite/<group>"])' \
+  2>&1 | tee /tmp/<branch>_<phase>.log
+grep -E "Error|Fail|Test Summary" /tmp/<branch>_<phase>.log
+
+# Full suite (mandatory for the last phase only)
+julia --project=@. -e 'using Pkg; Pkg.test()' \
+  2>&1 | tee /tmp/<branch>_final.log
+grep -E "Error|Fail|Test Summary" /tmp/<branch>_final.log
+```
+
+Never run the full suite for intermediate phases — it is too slow and hides errors.
 
 **Steps** — file-level granularity. "Edit `src/X/y.jl` to add method `f`" not "implement
 the feature". Concrete enough that a step is either done or not.
@@ -70,14 +113,23 @@ missed.
 
 **Docstrings** — always the last step of the last phase, once the API is stable.
 
+**Code snippets** — include short code snippets directly in the plan to guide the
+implementer. This applies to:
+
+- *Source changes*: show the before/after signature or struct definition.
+- *Call-site updates*: show the exact old and new lines when renaming symbols.
+- *Test patterns*: show the test structure (fake type + testset skeleton) when the
+  test pattern is non-obvious.
+Snippets should be *concise shapes*, not full implementations.
+
 **Out of scope** — explicit. Prevents scope creep and clarifies what the human should
 not expect after the plan is executed.
 
 ---
 
 ## Concrete example structure (abbreviated)
 
-```
+````markdown
 # Plan: rename content → dynamics trait
 
 ## What and why
@@ -97,23 +149,43 @@ names the axis. Purely mechanical, no behavior change.
 Steps:
 1. Rename src/Traits/content.jl → dynamics.jl; update all names inside.
 2. Update include path in src/Traits/Traits.jl; update exports.
-Checkpoint: test/suite/traits — all green.
+Checkpoint:
+
+```bash
+get_test_command test_args=["suite/traits"]   # MCP
+# fallback: julia --project=@. -e 'using Pkg; Pkg.test(; test_args=["suite/traits"])' 2>&1 | tee /tmp/rename_traits.log
+```
 
 ### Phase B — Adapt consumers (Configs, Solutions)
+
 Steps:
+
 1. Update src/Configs/* (content_trait → dynamics_trait, value names).
 2. Update src/Solutions/building.jl.
-Checkpoint: test/suite/configs + test/suite/solutions — all green.
+Checkpoint:
+
+```bash
+get_test_command test_args=["suite/configs", "suite/solutions"]   # MCP
+```
 
 ### Phase C — Tests + docs
+
 Steps:
+
 1. Rename and update test/suite/traits/test_content.jl → test_dynamics.jl.
 2. Update docs/api_reference.jl file list.
-Checkpoint: full suite green.
+Checkpoint (full suite — last phase):
+
+```bash
+get_test_command   # MCP, no test_args = full suite
+# fallback: julia --project=@. -e 'using Pkg; Pkg.test()' 2>&1 | tee /tmp/rename_final.log
+```
 
 ## Human checkpoints
+
 - ⛔ Ask before commit after Phase C.
 
 ## Out of scope
+
 Parametrizing AbstractSystem/AbstractFlow by the dynamics trait (see action_plan.md Phase C/D).
-```
+````

src/Descriptions/complete.jl

@@ -45,7 +45,7 @@ ERROR: AmbiguousDescription: the description (:f,) is ambiguous / incorrect
        Context: description completion
 ```
 
-See also: [`CTBase.Descriptions.compute_similarity`](@ref), [`CTBase.Descriptions.find_similar_descriptions`](@ref), [`CTBase.Descriptions.format_description_candidates`](@ref), [`CTBase.Exceptions.AmbiguousDescription`](@ref)
+See also: [`CTBase.Descriptions._compute_similarity`](@ref), [`CTBase.Descriptions._find_similar_descriptions`](@ref), [`CTBase.Descriptions._format_description_candidates`](@ref), [`CTBase.Exceptions.AmbiguousDescription`](@ref)
 """
 function complete(list::Symbol...; descriptions::Tuple{Vararg{Description}})::Description
     n = length(descriptions)
@@ -70,8 +70,8 @@ function complete(list::Symbol...; descriptions::Tuple{Vararg{Description}})::De
 
     if maximum(table[:, 2]) == 0
         # Find similar descriptions for helpful suggestions
-        similar_descs = find_similar_descriptions(list, descriptions; max_results=5)
-        all_candidates = format_description_candidates(descriptions; max_show=10)
+        similar_descs = _find_similar_descriptions(list, descriptions; max_results=5)
+        all_candidates = _format_description_candidates(descriptions; max_show=10)
 
         # Build contextual suggestion
         suggestion = if !isempty(similar_descs)

src/Descriptions/similarity.jl

@@ -23,7 +23,7 @@ julia> CTBase.Descriptions.compute_similarity((:x, :y), (:a, :b))
 0.0
 ```
 """
-function compute_similarity(desc1::Description, desc2::Description)::Float64
+function _compute_similarity(desc1::Description, desc2::Description)::Float64
     if isempty(desc1) || isempty(desc2)
         return 0.0
     end
@@ -62,7 +62,7 @@ julia> CTBase.Descriptions.find_similar_descriptions((:a,), descriptions)
  "(:a, :c)"
 ```
 """
-function find_similar_descriptions(
+function _find_similar_descriptions(
     target::Tuple{Vararg{Symbol}},
     descriptions::Tuple{Vararg{Description}};
     max_results::Int=5,
@@ -73,7 +73,7 @@ function find_similar_descriptions(
 
     # Compute similarities
     similarities = [
-        (compute_similarity(target, desc), string(desc)) for desc in descriptions
+        (_compute_similarity(target, desc), string(desc)) for desc in descriptions
     ]
 
     # Sort by similarity (descending) and take top results
@@ -118,7 +118,7 @@ julia> CTBase.Descriptions.format_description_candidates(descriptions; max_show=
  "(:x, :y)"
 ```
 """
-function format_description_candidates(
+function _format_description_candidates(
     descriptions::Tuple{Vararg{Description}}; max_show::Int=5
 )::Vector{String}
     if isempty(descriptions)

src/Exceptions/display.jl

@@ -44,7 +44,7 @@ Filters out Julia stdlib.
 # Returns
 - `Vector`: Filtered stacktrace frames
 """
-function extract_user_frames(st::Vector)
+function _extract_user_frames(st::Vector)
     user_frames = filter(st) do frame
         file_str = string(frame.file)
         # Keep frames that are NOT from Julia stdlib or exception display internals
@@ -68,7 +68,7 @@ Display an error in a user-friendly format with clear sections and user code loc
 - `io::IO`: Output stream
 - `e::CTException`: The exception to display
 """
-function format_user_friendly_error(io::IO, e::CTException)
+function _format_user_friendly_error(io::IO, e::CTException)
     #println(io, "\n" * "━"^70)
     _print_ansi_styled(io, "Control Toolbox Error\n", :red, true)
     #println(io, "─"^28)
@@ -232,7 +232,7 @@ function format_user_friendly_error(io::IO, e::CTException)
     end
 
     # Add user code location
-    user_frames = extract_user_frames(stacktrace(catch_backtrace()))
+    user_frames = _extract_user_frames(stacktrace(catch_backtrace()))
     if !isempty(user_frames)
         println(io, "📍 In your code:")
         # Show up to 3 most relevant user frames
@@ -262,7 +262,7 @@ Custom error display for IncorrectArgument.
 Shows user-friendly format with enriched information.
 """
 function Base.showerror(io::IO, e::IncorrectArgument)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -271,7 +271,7 @@ end
 Custom error display for PreconditionError.
 """
 function Base.showerror(io::IO, e::PreconditionError)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -280,7 +280,7 @@ end
 Custom error display for NotImplemented.
 """
 function Base.showerror(io::IO, e::NotImplemented)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -289,7 +289,7 @@ end
 Custom error display for ParsingError.
 """
 function Base.showerror(io::IO, e::ParsingError)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -298,7 +298,7 @@ end
 Custom error display for AmbiguousDescription.
 """
 function Base.showerror(io::IO, e::AmbiguousDescription)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -307,7 +307,7 @@ end
 Custom error display for ExtensionError.
 """
 function Base.showerror(io::IO, e::ExtensionError)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end
 
 """
@@ -316,5 +316,5 @@ end
 Custom error display for SolverFailure.
 """
 function Base.showerror(io::IO, e::SolverFailure)
-    return format_user_friendly_error(io, e)
+    return _format_user_friendly_error(io, e)
 end

src/Extensions/Extensions.jl

@@ -332,7 +332,10 @@ function run_tests(; kwargs...)
     return run_tests(TestRunnerTag(); kwargs...)
 end
 
-# Export public API (only user-facing functions, tags are internal)
+# Export public API
 export automatic_reference_documentation, postprocess_coverage, run_tests
+export AbstractDocumenterReferenceTag, DocumenterReferenceTag
+export AbstractCoveragePostprocessingTag, CoveragePostprocessingTag
+export AbstractTestRunnerTag, TestRunnerTag
 
 end # module