docs/result-template.md

DecBench C++ target validation result

Copy this file to results/<target>.md and replace every TBD with measured data. Do not infer or estimate collision statistics.

Target metadata

Field	Value
Target	TBD
Upstream	TBD
Release/tag	TBD
Resolved commit	TBD
Track	GCC/DWARF or MSVC/PDB
DecBench revision	d9f4f8af6097d7c42c4965cfc3f197dcf76f0a4f
Validation date	TBD
Host	TBD
Container / OS	TBD
Compiler	TBD
Linker	TBD
Windows SDK	N/A or TBD
Wine/msvc-wine	N/A or TBD

Build and ground-truth summary

Mode	Build	Linked image(s)	.ii count	Ground truth	Source-owned function addresses	Unique short names	Collision groups	Collision addresses	Collision rate
O0	TBD	TBD	TBD / N/A	DWARF/PDB: TBD	TBD	TBD	TBD	TBD	TBD
O2	TBD	TBD	TBD / N/A	DWARF/PDB: TBD	TBD	TBD	TBD	TBD	TBD
O2-noinline	TBD	TBD	TBD / N/A	DWARF/PDB: TBD	TBD	TBD	TBD	TBD	TBD

Collision rate is:

collision_addresses / source_owned_function_addresses

where collision_addresses counts distinct source-owned addresses/RVAs belonging to a short-name group with more than one distinct address.

Optimization control

Record the actual final compile/link flags, not only the intended configuration.

O0

compile: TBD
link:    TBD

O2

compile: TBD
link:    TBD

O2-noinline

compile: TBD
link:    TBD

Unexpected optimization/LTO/IPO/inlining behavior:

TBD

Linked images

O0

TBD

O2

TBD

O2-noinline

TBD

For PE targets, record the exact PE ↔ PDB pair.

Source ownership filter

Included project-owned paths/compilands:

TBD

Excluded tests/vendor/compiler-probe/generated paths:

TBD

Any uncertain ownership cases:

TBD

Short-name collision details

List the highest-impact collision groups. Keep qualified names and addresses/RVAs so the result can be re-audited.

Short name	Distinct addresses/RVAs	Example qualified names	Notes
TBD	TBD	TBD	TBD

Preprocessed source / oracle notes

For GCC/DWARF targets:

• .ii preservation: TBD
• DWARF DW_AT_specification handling checked: TBD
• DWARF DW_AT_abstract_origin handling checked: TBD

For MSVC/PDB targets:

• PDB generated for exact image: TBD
• source/compiland information available: TBD
• procedure/RVA extraction method: TBD

Final status

Choose one:

SELECTED      # source/build shape reviewed, measurements incomplete
VALIDATED     # all required local build + oracle + collision checks completed
CONFIRMED     # accepted as a clean/appropriate initial corpus target after review
REJECTED      # unsuitable; explain why below

Status: TBD

Decision rationale:

TBD

Remaining blockers:

TBD