Natvis complexity limit is enforced during expansion

[besuns-havok]

Type: Bug

The C++ Expression Evaluator will enforce the natvis complexity limit during expansion, at least for trees (ex: std::map). So any collection with a non-trivial number of elements will not format correctly if they key and/or value of the collection has an associated natvis rule.

A similar bug has been reported and fixed in Visual Studio, but the fix has not been applied to the cpptools extension in vscode.

Repro steps:

natvisrepro.zip

1. Unzip and open the attached folder.
2. Debug with the default configuration.
3. When the debugger stops at the __debugbreak() call, expand foo in the variables view.

Expected result: 100 elements are shown, each having a key and value.

Actual result: Only ~50 keys are shown, and no values are shown. Missing keys are shown as [] and missing values shown as (, ).

Extension version: 1.31.4
VS Code version: Code 1.116.0 (560a9dba96f961efea7b1612916f89e5d5d4d679, 2026-04-15T00:28:13Z)
OS version: Windows_NT x64 10.0.26100
Modes:

System Info

Item	Value
CPUs	13th Gen Intel(R) Core(TM) i9-13950HX (32 x 2419)
GPU Status	2d_canvas: enabled GPU0: VENDOR= 0x10de, DEVICE=0x2717 [NVIDIA GeForce RTX 4090 Laptop GPU], DRIVER_VENDOR=NVIDIA, DRIVER_VERSION=32.0.15.8160 ACTIVE GPU1: VENDOR= 0x1414, DEVICE=0x008c [Microsoft Basic Render Driver], DRIVER_VERSION=10.0.26100.7309 Machine model name: Machine model version: direct_rendering_display_compositor: disabled_off_ok gpu_compositing: enabled multiple_raster_threads: enabled_on opengl: enabled_on rasterization: enabled raw_draw: disabled_off_ok skia_graphite: disabled_off trees_in_viz: disabled_off video_decode: enabled video_encode: enabled webgl: enabled webgl2: enabled webgpu: enabled webnn: disabled_off
Load (avg)	undefined
Memory (System)	63.77GB (36.98GB free)
Process Argv
Screen Reader	no
VM	0%

[sean-mcmanus]

@besuns-havok This seems like the same issue as #14362. That linked VS bug appears to have been fixed a year ago, so we would have the fix already, so this current issue might be a new regression? It's also possible there might be some additional fix that is needed in the VS Code scenario.

[besuns-havok]

I concur, this issue is a duplicate of #14362.