net/http/internal — validation proof
Go’s own net/http/internal test suite, converted to C# by go2cs, built against the converted standard
library, run under the Go-semantics test host, and compared verdict for verdict against a clean
go test -json baseline of the same sources. This page is generated by the converter from that
comparison — it is the evidence behind the net/http/internal row in
Validated Test Packages.
Validated 2026-09-23 · converter f95f88866
14 matched · 1 disclosed — Go 1.24.13, windows/amd64, converted package
src/core/net/http/internal.
Measured at Release (tiered JIT off), oracle go version go1.24.13 windows/amd64.
Verdicts
| Test | go test |
go2cs |
|---|---|---|
TestChunk |
pass | pass |
TestChunkEndReadError |
pass | pass |
TestChunkInvalidInputs |
pass | pass |
TestChunkInvalidInputs/bare_LF_in_chunk_data |
pass | pass |
TestChunkInvalidInputs/bare_LF_in_chunk_extension |
pass | pass |
TestChunkInvalidInputs/bare_LF_in_chunk_size |
pass | pass |
TestChunkInvalidInputs/extra_LF_in_chunk_size |
pass | pass |
TestChunkReadMultiple |
pass | pass |
TestChunkReadPartial |
pass | pass |
TestChunkReaderAllocs |
pass | fail (disclosed) |
TestChunkReaderByteAtATime |
pass | pass |
TestChunkReaderTooMuchOverhead |
pass | pass |
TestChunkReadingIgnoresExtensions |
pass | pass |
TestIncompleteChunk |
pass | pass |
TestParseHexUint |
pass | pass |
Disclosed divergences
A disclosed divergence is a specific Go assertion this conversion does not satisfy — not
a skipped test and not a tolerance. Each one is pinned by exact failure signature in the package’s
hand-owned go2cs_test_disclosures.json;
a disclosed test that fails any other way is still a hard mismatch.
The Class column says which kind each one is: a deferred entry is an assertion the managed
CLR can meet, pinned against the named plan that will retire it; every other class is one it
provably cannot satisfy.
| Test | Class | Pinned reason |
|---|---|---|
TestChunkReaderAllocs |
deferred |
exact-count AllocsPerRun assert (fails above 1.5, i.e. wants 1): the measured loop constructs a fresh reader every iteration, and the converted NewChunkedReader emits TWO managed objects where Go allocates one - the box Ꮡ(new chunkedReader(r: br)), which is the direct analogue of Go’s &chunkedReader{}, plus a chunkedReaderжReader interface shell, because C# has no two-word interface value and the shell IS the itab+data fused into an object (docs/phase4/DESIGN-iface-shell-caching.md section 2). Measured 2 objects/run against Go’s 1, deterministically. Not amortizable for this shape: P3’s per-referent shell cache is keyed on the ж-box and this loop mints a new reader each iteration, so a fresh shell is required regardless - the same doc rules that nothing removes the shell allocation short of IDynamicInterfaceCastable. Distinct from alloc-count-semantics: go2cs’s counter is a true object count here (a lower bound on the real total), not an incomparable byte-derived shim, so the divergence is a genuine profile difference and is disclosed as one. RELABEL 2026-09-23 (C1, as ruled at ledger 2026-09-23 03:37 O1): alloc-profile -> deferred. The 2026-08-25 ATTRIBUTION above is SUPERSEDED: the two counted objects are the box at chunked.cs:33 (the direct analogue of Go’s &chunkedReader{}, Go’s own allocation) and chunkedReader.buf, an array<byte> buf = new(2) field (chunked.cs:40), which is the excess. The chunkedReaderжReader shell at :33 and the per-read bufio shell at :114 are UNCOUNTED (AllocationCounter.cs:50-53). The shell’s structural ruling itself stands. |