 |
OpenSWMM Engine
6.0.0-alpha.1
Data-oriented, plugin-extensible SWMM Engine (6.0.0-alpha.1)
|
|
OpenSWMM Engine
6.0.0-alpha.1
Data-oriented, plugin-extensible SWMM Engine (6.0.0-alpha.1)
|
Relational (normalized) Structure-of-Arrays side-tables for link subtypes. More...
#include <algorithm>#include <vector>#include <cstdint>#include <string>#include <type_traits>#include "LinkData.hpp"Go to the source code of this file.
Classes | |
| struct | openswmm::ConduitData |
| Dense SoA for conduit-only properties (one row per CONDUIT link). More... | |
| struct | openswmm::PumpData |
| struct | openswmm::OrificeData |
| struct | openswmm::WeirData |
| struct | openswmm::OutletData |
| struct | openswmm::LinkSubtypes |
| Owns the five link subtype side-tables plus the reverse index map. More... | |
Namespaces | |
| namespace | openswmm |
Relational (normalized) Structure-of-Arrays side-tables for link subtypes.
Phase 6 of the relational refactor — the link analogue of NodeSubtypes.hpp. The wide LinkData SoA allocates every conduit/pump/orifice/weir/outlet field for all links even though each field is only valid when type[i] matches. This header holds dense per-subtype side-tables — one per link type — sized to that type's count and joined back to the base link by a stored link_idx ("foreign key"). A reverse map (subtype_row) gives O(1) base-index → row lookup.
Authoritative: the side-tables are the single source of truth for subtype config. Parse/edit writers populate the rows directly; the StructureSolver / XSectGroups init source-reads and the C-API read them. Rows are kept in ascending link_idx order so per-row iteration matches a base-link-ascending scan bit-for-bit.
Design note — shared fields stay on base LinkData: the cross section (xsect_*), has_flap_gate, and pump_curve_name are used by more than one link type (xsect/has_flap by conduit+orifice+weir; pump_curve_name by pump+outlet+conduit-IRREGULAR), so they remain on the base LinkData rather than any one subtype table. Only the unambiguously type-specific fields move here. Structures are split into Orifice/Weir/Outlet with NAMED columns over the type-overloaded wide param1/param2 discriminators (stored as double to preserve bit-for-bit parity with the legacy compute).
1.13.2