[jogibear9988]

A member assignment value containing try/finally can invalidate the pending initializer receiver. Keep the constructed object and assignment receiver in locals while the value is evaluated, then reload the receiver and value for assignment. Record the affected initializer and RHS nodes during collection in existing complex-argument metadata, so an unrelated try does not activate spilling.

Class and struct field/property assignments, result values and left/right/finally effects, compared with System compilation. On .NET Framework the struct EH initializer uses equivalent direct C# as its oracle because the Framework System compiler rejects that shape; class cases use System compilation. 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	489 passed	1,756 passed
.NET 8	489 passed	1,755 passed
.NET Framework 4.7.2	468 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	1121.5 → 1100.7	1.9 → 1.9	872.0 → 872.0	0.0 → 0.0
Standard / member_init	1520.6 → 1703.8	3.8 → 3.7	1064.1 → 1064.1	24.0 → 24.0
Standard / member_init_unrelated_try	1954.8 → 2031.0	4.0 → 4.3	1664.1 → 1664.1	24.0 → 24.0
Light / add_control	1058.1 → 1144.7	2.0 → 2.0	872.0 → 872.0	0.0 → 0.0
Light / member_init	1251.1 → 1327.5	3.5 → 4.0	1064.0 → 1064.0	24.0 → 24.0
Light / member_init_unrelated_try	1895.5 → 1673.6	3.6 → 3.7	1664.0 → 1664.0	24.0 → 24.0

net10.0 results (master → change)

API / workload	Compile ns	Invoke ns	Compile B/op	Invoke B/op
Standard / add_control	5567.0 → 7312.5	4.1 → 4.1	872.1 → 872.1	0.0 → 0.0
Standard / member_init	5366.1 → 5381.5	6.2 → 6.1	1064.3 → 1064.3	24.0 → 24.0
Standard / member_init_unrelated_try	6088.3 → 6220.2	6.8 → 6.8	1664.3 → 1664.3	24.0 → 24.0
Light / add_control	3559.9 → 3226.7	2.2 → 2.3	872.0 → 872.0	0.0 → 0.0
Light / member_init	3712.2 → 3352.5	4.4 → 4.2	1064.0 → 1064.0	24.0 → 24.0
Light / member_init_unrelated_try	3931.1 → 4111.3	4.5 → 4.4	1664.0 → 1664.0	24.0 → 24.0

Emitted IL differs in these measured workloads: none.
No measured allocation increase above 0.1 B/op in these controls.