External OSS Attribution: Two Executed Repro Cards - 2026-06-15
External OSS Attribution: Two Executed Repro Cards - 2026-06-15
This artifact is not a claim that Mimesis becomes correct by citing famous code.
It is the opposite:
Strong external OSS can expose load-bearing structure, but attribution is not validation transfer.
The useful public claim is narrower and stronger:
Digital Factory has two executed local repro cards showing how selected external
OSS artifacts can expose load-bearing structure under objective oracles.
This is source-level attribution under reproduction, not downstream lift, not
maintainer endorsement, not legal clearance, and not external validation.
Claim
Allowed public claim:
Mimesis can use selected external OSS artifacts as attribution anchors for
load-bearing structure when a local repro card includes a source pointer,
objective oracle, defect control, wrong-anchor control, command transcript,
and forbidden-claim boundary.
Forbidden public claim:
- external OSS validation transfers to Mimesis outputs,
- source-level attribution is downstream lift,
- Go or Rust maintainers endorsed Mimesis,
- the repro cards are legal clearance,
- the repro cards prove production readiness,
- the two examples prove universal method validity,
- or the extracted structure is safe to market without downstream gates.
Verified Originals
These sources are attribution anchors. They do not validate Mimesis Engineering.
| source | what it contributes |
|---|---|
| golang/go issue 67470 | Concrete Go time.Parse defect report for out-of-range timezone offsets such as +99:99, plus the repair shape: reject impossible offset hour/minute fields instead of silently constructing a broken offset. |
| Go time parser source | Parser structure and timezone offset validation context for Parse, ParseInLocation, and the internal parser functions. |
| Go RFC3339 parser source | Fixed-width range-guard style for date, time, and timezone fields. |
| Go strconv docs | The public contract that ParseFloat returns the nearest floating-point number under IEEE 754 unbiased rounding. |
| Go Eisel-Lemire source | Fast-path float parsing structure that returns to a slower exact path on half-way ambiguity. |
| Rust PR 86761 | Prior-art context for Eisel-Lemire / decimal-to-float half-way handling in a production language runtime. |
| CPython strtod tests | Independent nearest-even parsing test tradition used as an oracle reference for float parsing behavior. |
| Number Parsing at a Gigabyte per Second | Original Eisel-Lemire algorithm paper lineage for fast decimal-to-binary floating-point conversion. |
Attribution Boundary
This packet records source-level attribution, not implementation ownership.
Safe wording:
Validated against Go standard-library parsing behavior and Eisel-Lemire /
Go strconv prior-art references for nearest-even float parsing.
Unsafe wording:
Because Go and Rust are verified, Mimesis output is verified.
Executed Repro Card A - P-EXT-03 / Go time.Parse
| field | value |
|---|---|
| Card | P-EXT-03 |
| Source anchor | golang/go issue 67470 plus Go time.Parse source context. |
| Load-bearing structure | Insert an explicit timezone offset range guard before constructing the parsed zone offset. |
| Objective oracle | Out-of-range timezone offsets such as +99:99 must error, not round-trip into a malformed formatted offset. |
| Defect control | A parser that accepts and constructs +99:99 fails the oracle. |
| Wrong-anchor control | A clamp/normalize-style repair fails because the oracle requires rejection, not repair. |
| Local result | Go standard library fix behavior passed, local fix-structure passed, defect failed, wrong-anchor failed. |
| Boundary | This proves a local reproduction of one parser-guard structure, not general parser correctness. |
Observed command sequence:
go version
go run parse_attribution.go
go build -o .\parse_attribution_local.exe parse_attribution.go
.\parse_attribution_local.exe
Observed environment and result summary:
go version go1.26.2 windows/amd64
go run temp-exe was blocked
go build -o local executable workaround succeeded
stdlib(FIX) PASS
fix-structure PASS
defect FAIL
wrong-anchor FAIL
The interesting part is not that Go exists. The interesting part is that the defect and wrong-anchor controls both fail under the same objective oracle.
Executed Repro Card B - P-EXT-06 / Eisel-Lemire
| field | value |
|---|---|
| Card | P-EXT-06 |
| Source anchor | Eisel-Lemire / Go strconv / Rust decimal-to-float prior-art references. |
| Load-bearing structure | Half-way ambiguity must not be guessed by the fast path; it must defer to exact nearest-even behavior. |
| Objective oracle | Python float() bit pattern comparison and nearest-even expectations for half-way cases. |
| Defect control | A no-guard, round-up implementation mis-rounds tie cases. |
| Wrong-anchor control | A round-half-to-odd policy produces distinct failures. |
| Local result | Fix passed, defect failed, wrong-anchor failed, controls passed. |
| Boundary | This proves one load-bearing rounding guard under local oracle tests, not universal float-parser conformance. |
Observed command:
python -B p_ext_06_repro.py
Observed result summary:
DEFECT (no guard, round-up): FAIL(2)
FIX (guard->round-half-even): PASS
WRONG_ANCHOR (round-half-to-ODD): FAIL(3)
VERDICT: REPRODUCTION CONFIRMED -- halfway guard is load-bearing
The phrase that matters from the Go public contract is simple: ParseFloat returns the nearest floating-point number. The local card uses that kind of oracle boundary: when the value is halfway, the fast attractive shortcut must not invent a different rounding policy.
Commands / Redlines
Local workbench verification commands observed after the repro cards:
python -B verify_workbench_surface.py
cd mimesis-plugin
python -B verify_evidence_references.py
python -B tools\validate_module.py --all
Observed result:
Digital Factory workbench surface checks passed.
Mimesis evidence-reference checks passed.
14/14 valid
Redlines:
- no source anchor without a local oracle,
- no local oracle without defect and wrong-anchor controls,
- no reproduction claim from file presence,
- no inherited validation claim,
- no claim that source attribution is downstream lift,
- no maintainer endorsement claim,
- no legal clearance claim.
Allowed Claim
Safe sentence:
Mimesis external OSS attribution now has two executed repro cards: Go
time.Parse timezone range guarding and Eisel-Lemire nearest-even float parsing.
Both cards keep the source anchor, objective oracle, defect control,
wrong-anchor control, command result, and forbidden claims together.
Forbidden Claim
Unsafe sentence:
Mimesis is validated because it is based on Go, Rust, and Eisel-Lemire.
This is specifically forbidden because source-level attribution is not downstream lift, not maintainer endorsement, not legal clearance, and not external validation.
Claim Boundary
What this artifact proves:
- two local external OSS attribution repro cards were executed,
P-EXT-03records a Gotime.Parsetimezone range-guard reproduction,P-EXT-06records an Eisel-Lemire / nearest-even float parsing reproduction,- both cards use an objective oracle plus defect and wrong-anchor controls,
- and both cards preserve forbidden claims beside the evidence.
What this artifact does not prove:
- external validation,
- downstream output improvement,
- production readiness,
- universal method validity,
- legal clearance,
- maintainer endorsement,
- full parser conformance,
- or public adoption.
Next Proof
The next stronger artifact is not more dramatic language.
It should be either:
- a public-safe casebook card that turns these two repro cards into a reusable worksheet,
- a third repro card in a different source domain with the same oracle/control structure,
- or one downstream case where this attribution structure changes a target output and then passes a separate downstream gate.
Until then, this remains a proof-bounded external OSS attribution packet, not external validation.