Python API: edge neighborhood nodes retain visitors after leaving scope

Abierto
  • ¿Algo no está bien?
#2,542 1 comentario 0 reacciones 1 asignado Ver en GitHub

Los mantenedores suelen responder en 1 día

@klayoutmatthias ya está trabajando en esto.

Desde el 10/11/2026.

  • #2543 de @klayoutmatthias — abierto

Evaluación

Este issue todavía no se ha evaluado.

Descripción

bug

Creating CompoundRegionOperationNode.new_edge_neighborhood leaves its Python visitor alive after the node wrapper and all Python references to the visitor have been dropped. Repeated construction accumulates visitors and any data attached to them. Calling Region.complex_op is not needed to reproduce this.

Expected: once the node and visitor are no longer referenced, the visitor can be collected.

Minimal reproduction:

import gc
import weakref
from typing import List, Tuple
import klayout.db as kdb

class Visitor(kdb.EdgeNeighborhoodVisitor):
    def on_edge(self, layout: kdb.Layout, cell: kdb.Cell, edge: kdb.Edge, neighborhood: List):
        pass

def make() -> Tuple[weakref.ReferenceType, weakref.ReferenceType]:
    visitor = Visitor()
    node = kdb.CompoundRegionOperationNode.new_edge_neighborhood(
        children=[], visitor=visitor)
    return weakref.ref(visitor), weakref.ref(node)

references = [make() for _ in range(100)]
gc.collect()
print("Visitors alive:", sum(visitor() is not None for visitor, _ in references))
print("Node wrappers alive:", sum(node() is not None for _, node in references))

Actual: 100 visitors and zero node wrappers remain alive.

Controls:

  • Constructing and discarding visitors without constructing nodes leaves zero of 100 visitors alive
  • Creating nodes without executing them leaves 1, 10 and 100 visitors alive after 1, 10 and 100 calls
  • Executing complex_op on ordinary flat rectangular regions gives the same 1, 10 and 100 counts
  • The node-only and flat-region reproductions also retain all 100 visitors on Linux with Python 3.12.3

The reproduction has no extraction context, DeepShapeStore, callback instrumentation or retained strong-reference collection. Its observation list contains only weak references.

This is distinct from #2495, which concerned retained on_edge arguments and was fixed by #2501. Here no callback needs to run. In KPEX, each visitor refers to the extraction results, so retained visitors can keep completed extraction data alive when operations are repeated in one process.

The v0.30.13 source appears consistent with this lifetime observation:

Versions:

  • KLayout Python module 0.30.13, Python 3.14.6, macOS 26.6.2
  • Independent node-only and flat-region reproductions: KLayout Python module 0.30.13, Python 3.12.3, Linux
  • The KLayout executable is not involved
  • Source tag v0.30.13, commit 06b8ff4909ad1696c06c634183d8fd0eaa8a1b83
Lenguaje dominante
C++
Estrellas
1.2k
Forks
305
Merge medio
3 d 2 h
PR fusionados (30 d)
11

Preparar el entorno

Este proyecto no incluye contenedor de desarrollo, Dockerfile ni guía de contribución, así que la configuración corre por tu cuenta: empieza por su README y consulta nuestra guía para la primera contribución para los pasos generales.

Primeros pasos

  1. Lee el issue completo y luego la guía de contribución del proyecto.
  2. Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
  3. Haz un fork del repositorio y trabaja en una rama.
  4. Abre un pull request que haga referencia al número del issue.

Más de KLayout/klayout

Todos los issues de KLayout/klayout

Issues similares

Más issues de C++

Recibe los nuevos issues en tu correo

Un resumen breve de issues de GitHub para principiantes.