[jogibear9988]

Collection Add arguments containing exception regions can clear the collection receiver and earlier argument values from the stack. Save the collection and Add arguments before reloading them in call order, reusing the existing constructor emitter. Record actual exception-region arguments during collection and spill only the affected collection expression and Add calls.

Class and struct collections, multiple Add arguments, cleanup, throwing arguments, nonvoid Add results, ignored results and constructor arguments. Regressions require strict FEC compilation. Native suites cover Standard and LightExpression.

Validation

Local validation uses master baseline 094d067 plus this change. Release solution build: zero warnings/errors.

Runtime	TestX groups	Legacy tests
.NET 10	495 passed	1,756 passed
.NET 8	495 passed	1,755 passed
.NET Framework 4.7.2	474 passed	1,639 passed

Measurements

Master baseline 094d0678 → this change, Release, Windows ARM64, .NET 8.0.28 / 10.0.9. Compilation uses prebuilt trees; runtime measures invocation through the same wrapper. Two fresh process rounds with reversed load order, seven samples per round, about 30 ms calibration, one logical CPU and tiered compilation disabled. Tables use the median of the two per-round medians.

The controls below produce correct results on master for their measured inputs. Newly supported trees have no valid before/after execution comparison because master rejects them, produces incorrect results or emits invalid IL. Timings include machine and JIT/code-placement variation; the unchanged add control helps show that variation. These observations do not prove universal performance neutrality.

net8.0 results (master → change)

API / workload	Compile ns	Invoke ns	Compile B/op	Invoke B/op
Standard / add_control	1084.2 → 1152.6	2.0 → 2.0	872.0 → 872.0	0.0 → 0.0
Standard / list_initializer	1383.7 → 1355.5	11.1 → 12.5	1008.0 → 1008.0	72.0 → 72.0
Standard / list_initializer_unrelated_try	1933.9 → 1793.2	11.7 → 10.5	1608.0 → 1608.0	72.0 → 72.0
Light / add_control	1151.1 → 1137.2	2.0 → 2.0	872.0 → 872.0	0.0 → 0.0
Light / list_initializer	1219.9 → 1334.8	10.8 → 10.3	1040.0 → 1008.0	72.0 → 72.0
Light / list_initializer_unrelated_try	1623.5 → 1793.9	13.1 → 13.9	1640.0 → 1608.0	72.0 → 72.0

net10.0 results (master → change)

API / workload	Compile ns	Invoke ns	Compile B/op	Invoke B/op
Standard / add_control	4184.8 → 4213.3	2.0 → 1.9	872.1 → 872.1	0.0 → 0.0
Standard / list_initializer	3377.3 → 3665.1	38.8 → 43.9	1008.0 → 1008.0	72.0 → 72.0
Standard / list_initializer_unrelated_try	4144.3 → 4006.0	44.1 → 41.3	1608.0 → 1608.0	72.0 → 72.0
Light / add_control	4543.6 → 4109.4	2.0 → 2.0	872.0 → 872.0	0.0 → 0.0
Light / list_initializer	2959.2 → 3235.3	44.6 → 45.4	1040.0 → 1008.0	72.0 → 72.0
Light / list_initializer_unrelated_try	3282.6 → 3382.2	42.3 → 41.0	1640.0 → 1608.0	72.0 → 72.0

Emitted IL differs in these measured workloads: none.
No measured allocation increase above 0.1 B/op in these controls.
Light / list_initializer compile allocation: 32 B/op less than master.
Light / list_initializer_unrelated_try compile allocation: 32 B/op less than master.