feat(#5525): file-local name handles for anonymous >> objects via >> foo

[yegor256]

Closes #5525.

What

Extends the >> auto-name suffix to optionally carry a trailing NAME — a file-local handle:

[] > app
  [n] >> fibo
    if. > @
      n.lt 2
      n
      plus.
        fibo (n.minus 1)
        fibo (n.minus 2)
  stdout > @
    sprintf "%d is the %dth number\n" *(fibo 6) 6

>> fibo keeps the object anonymous — it still receives its deterministic cactus @name (§9.2) and never enters the visible namespace — but fibo becomes a typeable alias for that cactus name, usable anywhere in the same .eo file. Here fibo is referenced twice recursively inside its own body and once from app's @, and all three rewrite to the same cactus @name. Unlike plain > fibo, the object is never put on app's public surface. Pure syntactic sugar — no new semantics.

This supersedes the reverted % (#5433): % self-referenced only the nearest enclosing anonymous formation, whereas >> foo gives a typeable handle to any anonymous object reachable from anywhere the cactus name is in scope (self-recursion, mutual recursion, sibling calls).

How

Mirrors the marker + first-pass-XSL pattern of #5432 / #5522.

• Parser front-end. Suffix.auto() reads an optional handle after >> (and the ! const), surfaced via the new Suffix.handle(). Emit.local() records it as an @local marker on the object that carries the cactus @name; it is emitted by LnFormation, LnOnlyPhi and ChainEmission, so >> name is accepted uniformly on bare formations, inline-phi formations and applications. ! const stays allowed (>>! name); /sig stays forbidden (R-3.10.2).
• New first-pass resolve-local-names.xsl, wired into Canonical right after resolve-self.xsl, before build-fqns.xsl / scope resolution. It (1) collects the per-file @local → @name table, (2) rewrites every @base equal to a handle into the matching cactus @name, (3) drops the @local markers, and (4) reports a duplicate handle as a resolve-local-names compile-time error (duplicate local name 'name'). A reference with no matching handle is left untouched for later scope resolution. Downstream passes see only ordinary references by the reserved cactus name — nothing changes below the parser front-end.
• Spec. PARSER_SPEC.md gains R-3.10.12 (file-local handles) and R-9.2.3 (emission + rewrite), plus the >> base-form table entry.

Tests

• eo-packs/parse/resolve-local-names.yaml — the reshape in isolation: three fibo references rewrite to the single cactus @name, @local markers dropped, no errors.
• eo-packs/parse/resolve-local-names-duplicate.yaml — two >> dup in one file produce the resolve-local-names duplicate-handle error.
• SuffixTest — >> fibo yields Form.AUTO carrying the handle; >>! fibo combines const with the handle; trailing garbage after the handle is still rejected (trailing-non-suffix-auto typo pack updated).

All eo-parser tests pass (959, incl. the parse packs); qulice, xcop and yamllint are clean. I also verified the full parse train on the Fibonacci example end-to-end: all fibo references resolve to the single cactus @name with zero errors — identical to the resolved XMIR of the existing snippets/fibo.yaml (which uses > f), so runtime behaviour is unchanged.

The diff is kept under the 200-line pr-size limit (197). To stay within it, the runtime SnippetIT snippet was deferred to a follow-up — its program is the existing snippets/fibo.yaml with > f → >> fibo, and the resolution is already covered by the isolation packs and the full-train check above. Happy to add it (or split the duplicate-handle safeguard out) if you'd prefer a different balance.

🤖 Generated with Claude Code

---

Generated by Claude Code

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[github-actions]

⚠️ Benchmark Comparison Unavailable

Unfortunately, one of the benchmarks is missing, and we couldn't generate a performance comparison report.

Please ensure that both the base and PR benchmark results are available for analysis.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[0crat]

@yegor256 Thanks for the contribution! You've earned +4 points for this: +16 as a basis; -4 for too many hits-of-code (197 >= 100); -8 for the lack of code review. Please, keep them coming. Your running score is +1808; don't forget to check your Zerocracy account too).