What

Removes the reflection-based pointer-form fallback from SubjectAs[T] (and the reflect import), making SubjectAs/SET.IssSub match the value form of a sub_id Subject Identifier only.

Why

Before go-subjectid v0.2.0, Parse returned the pointer form while a Go-built SET held the value form, so SubjectAs carried a reflection branch to absorb both. v0.2.0 (#17) made Parse return the value form — the same shape a Go-built SET holds — so no supported flow ever places a pointer in SET.Subject. The reflection branch is dead for every supported input.

SubjectAs is now nil guards + a single s.Subject.(T) value assertion.

Behavior change

A SET hand-built with a *subjectid.IssSubID Subject now reports ok=false from SubjectAs/IssSub instead of being dereferenced. No supported flow produces that shape (Parse yields a value; a Go-built SET should hold a value), so consumers reading parsed subjects or building value-form SETs are unaffected. A typed-nil pointer Subject still returns ok=false without panicking — the value assertion fails without touching the pointee, which is what the old !rv.IsNil() guard protected.

Changes

• subject.go — drop the reflection block + reflect import; rewrite SubjectAs/IssSub godoc to drop the value/pointer-absorption framing.
• secevent.go — trim the SET.Subject field doc that described the pointer form as a possible held shape.
• example_test.go — trim ExampleSET_IssSub's value-or-pointer mention.
• subject_test.go — TestSubjectAsPointerForm (asserted pointer form IS read) → TestSubjectAsPointerFormUnmatched (asserts hand-built pointer form yields ok=false), pinning the value-only contract; the typed-nil subtest comment updated (no reflection path).
• encode_test.go — subject literal moves to value form for suite consistency.
• CHANGELOG.md — [Unreleased] → Changed entry.

Verification

Local CI mirror all green: gofmt -l, go vet ./..., go mod tidy -diff, go test -race -shuffle=on ./..., golangci-lint v2.12.2 run ./... (0 issues). Code review clean.

🤖 Generated with Claude Code