ConfigParam.hpp File Reference

pybind11-style configuration registration with macro-based field declaration and namespace-scoped tag kwargs. More...

#include "ConfigRegistry.hpp"
#include <type_traits>
#include <utility>

Include dependency graph for ConfigParam.hpp:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
struct	DNDS::ConfigTypeTagOf< T, Enable >
struct	DNDS::ConfigTypeTagOf< bool >
struct	DNDS::ConfigTypeTagOf< std::string >
struct	DNDS::ConfigTypeTagOf< T, std::enable_if_t< std::is_integral_v< T > &&!std::is_same_v< T, bool > > >
struct	DNDS::ConfigTypeTagOf< T, std::enable_if_t< std::is_floating_point_v< T > > >
struct	DNDS::ConfigTypeTagOf< std::vector< T > >
struct	DNDS::ConfigTypeTagOf< std::array< T, N > >
struct	DNDS::ConfigTypeTagOf< nlohmann::ordered_json >
struct	DNDS::ConfigTypeTagOf< T, std::enable_if_t< std::is_enum_v< T > > >
	C++ enum types serialize as JSON strings via nlohmann, so map to Enum. More...
struct	DNDS::detail::is_eigen_type< T, typename >
struct	DNDS::detail::is_eigen_type< T, std::void_t< typename T::Scalar, decltype(static_cast< int >(T::RowsAtCompileTime)), decltype(static_cast< int >(T::ColsAtCompileTime))> >
struct	DNDS::ConfigTypeTagOf< T, std::enable_if_t< detail::is_eigen_type< T >::value > >
struct	DNDS::Config::RangeTag
	Numeric range constraint. Enforced at parse time in readFromJson and emitted in schema. More...
struct	DNDS::Config::InfoTag
	Auxiliary info tag (emitted as "x-<key>" in schema). More...
struct	DNDS::Config::EnumValuesTag
	Explicit enum allowed-values tag. More...
class	DNDS::ConfigSectionBuilder< T >
	Builder object passed to the user's registration function. More...

Namespaces
namespace	DNDS
	the host side operators are provided as implemented
namespace	DNDS::detail
namespace	DNDS::Config
	Namespace for config field tag kwargs.

Macros
#define	DNDS_FIELD(name_, desc_, ...)    config.field(&T::name_, #name_, desc_, ##__VA_ARGS__)
	Register a field inside a DNDS_DECLARE_CONFIG body.
#define	DNDS_DECLARE_CONFIG(Type_)
	Open a config section registration body.

Functions
std::string	DNDS::schemaTypeString (ConfigTypeTag tag)
RangeTag	DNDS::Config::range (double min)
	Create a minimum-only range constraint.
RangeTag	DNDS::Config::range (double min, double max)
	Create a min+max range constraint.
InfoTag	DNDS::Config::info (std::string key, std::string value)
	Create an auxiliary info tag.
EnumValuesTag	DNDS::Config::enum_values (std::vector< std::string > vals)
	Create an enum allowed-values tag.
void	DNDS::detail::applyTag (FieldMeta &meta, const Config::RangeTag &tag)
void	DNDS::detail::applyTag (FieldMeta &meta, const Config::InfoTag &tag)
void	DNDS::detail::applyTag (FieldMeta &meta, const Config::EnumValuesTag &tag)
void	DNDS::detail::applyTags (FieldMeta &)
template<typename Tag , typename... Rest>
void	DNDS::detail::applyTags (FieldMeta &meta, Tag &&tag, Rest &&...rest)
template<typename T , typename V >
std::function< nlohmann::ordered_json()>	DNDS::detail::makeSchemaEntry (V T::*member, const char *desc, const FieldMeta &meta)
	Build the schemaEntry closure from a fully-tagged FieldMeta.
template<typename T , typename V >
std::function< void(const nlohmann::ordered_json &, const char *)>	DNDS::detail::makeRangeChecker (const FieldMeta &meta)
	Build a runtime range-check closure.

Detailed Description

pybind11-style configuration registration with macro-based field declaration and namespace-scoped tag kwargs.

Overview

Config sections are plain structs. Metadata is registered in a static method opened by the DNDS_DECLARE_CONFIG(Type) macro. Inside that body the user calls DNDS_FIELD(member, "description", tags...) which is a macro that auto-stringifies the member name — no name duplication.

Example

struct ImplicitCFLControl
{
    real CFL = 10;
    int  nForceLocalStartStep = INT_MAX;
    bool useLocalDt = true;
    real RANSRelax = 1;
 
    DNDS_DECLARE_CONFIG(ImplicitCFLControl)
    {
        DNDS_FIELD(CFL,                   "CFL for implicit local dt");
        DNDS_FIELD(nForceLocalStartStep,   "Step to force local dt",
                   DNDS::Config::range(0));
        DNDS_FIELD(useLocalDt,             "Use local (vs uniform) dTau");
        DNDS_FIELD(RANSRelax,              "RANS under-relaxation factor",
                   DNDS::Config::range(0.0, 1.0));
 
        config.check([](const T &s) -> DNDS::CheckResult {
            if (s.RANSRelax <= 0)
                return {false, "RANSRelax must be positive"};
            return {true, ""};
        });
    }
};

Adding a New Parameter

1. Declare the member with default (plain C++).
2. Add DNDS_FIELD(member, "description") in the DNDS_DECLARE_CONFIG body.

Tag kwargs (namespace-scoped)

Tags are passed as extra arguments to DNDS_FIELD:

Tag	Purpose	Example
DNDS::Config::range(min)	Min constraint (schema + runtime check)	DNDS::Config::range(0)
DNDS::Config::range(min,max)	Min+max constraint	DNDS::Config::range(0.0, 1.0)
DNDS::Config::info(k,v)	Aux info ("x-<key>" in schema)	DNDS::Config::info("unit","Pa")
DNDS::Config::enum_values(v)	Allowed string values for enum fields	DNDS::Config::enum_values({"Roe","HLLC"})

Runtime Range Validation

When a field has a range() tag, readFromJson() checks the parsed value against the min/max bounds and throws std::runtime_error with a clear message on violation. This catches bad config values at load time.

Section-Level Checks

config.check([](const T &s) -> DNDS::CheckResult { ... });
config.check_ctx([](const T &s, const DNDS::ConfigContext &ctx) -> DNDS::CheckResult { ... });

Nested Sections and Special Fields

Use explicit config.* calls (not the DNDS_FIELD macro) for these:

config.field_section(&T::frameRotation, "frameConstRotation", "Rotating frame");
config.field_array_of<BoxInit>(&T::boxInits, "boxInitializers", "Box inits");
config.field_map_of<CoarseCtrl>(&T::coarseList, "coarseGridList", "Per-level");
config.field_json(&T::extra, "odeSettingsExtra", "Opaque ODE settings");
config.field_alias(&T::rsType, "riemannSolverType", "Riemann solver type");

Trivial Copyability

The struct has no base class, no virtual methods, no instance-level data introduced by the macro. DNDS_DECLARE_CONFIG only generates static methods and friend functions.

Definition in file ConfigParam.hpp.

Macro Definition Documentation

DNDS_DECLARE_CONFIG

#define DNDS_DECLARE_CONFIG

(

Type_

)

Open a config section registration body.

Expands to:

• using T = Type (so DNDS_FIELD can reference &T::member).
• Lazy _dnds_ensure_registered() with one-time init guard.
• Friend to_json / from_json calling ensureRegistered first.
• schema(), validate(), validateWithContext(), validateKeys().
• Opens static void _dnds_do_register(ConfigSectionBuilder<T>& config) — the user provides the { ... } body after the macro.

struct MySection
{
    real a = 1.0;
    int  b = 42;
 
    DNDS_DECLARE_CONFIG(MySection)
    {
        DNDS_FIELD(a, "The a parameter", DNDS::Config::range(0));
        DNDS_FIELD(b, "The b parameter");
    }
};

Definition at line 662 of file ConfigParam.hpp.

DNDS_FIELD

#define DNDS_FIELD	(		name_,
			desc_,
			...
	)		config.field(&T::name_, #name_, desc_, ##__VA_ARGS__)

Register a field inside a DNDS_DECLARE_CONFIG body.

Auto-stringifies the member name so you never write it twice. The JSON key equals the C++ member name.

Parameters

name_	Member name (unquoted).
desc_	Description string literal.
...	Zero or more DNDS::Config tag objects.

DNDS_FIELD(CFL, "CFL for implicit dt", DNDS::Config::range(0));

Definition at line 633 of file ConfigParam.hpp.