Retire std.string_type String: delete dag/std/string_type.dag (0 importers after #12512)

[gunbai-bot]

Retires std.string_type String (= FreeMonoid<Char>), the duplicate of v2.std.text String. This is the operator-ruled retirement that gunbc#12512's import deletions made possible.

What changes

• dag/std/string_type.dag is deleted. Its only remaining declaration was type String, and an empty module is dead. Text crossings unfold or refuse: exact-representation text compat (XL-0T ruling B) #12512 had already deleted its two functions.
• v1.compiler.coercion structural_declaration_modules_for no longer lists the deleted path. The stage0 mirror is regenerated to a fixed point.
• test.claim.self_host_symbol_identity_binding_witness_test drops its decl_ref(std.string_type, String) sibling citation, because the cited-declaration gate refuses a citation of a declaration that does not exist.
• gunbc.defork_type_name_census: the String row moves from ForkAwaitingRuling to resolved. The resolution type had arms for deleting the v2 side but none for deleting the dag side, so this adds ResolvedByDeletingDag, the mirror of ResolvedByDeletingV2, and its render arm.
• v1.tests.claim.checkpoint_identity_keying_witness_test: two claims whose only subject was the deleted path are deleted. They sent dag/std/string_type.dag through the structural roster. Their v2.std.text siblings still cover the structural-String refusal. This file is in the v1 tests tree that the floor does not execute; its stage0 mirror is compiled by the lint step.

Zero importers, verified

• Multi-line-aware search at this head: git grep -lzE 'import[[:space:]]+std\.string_type\b'. It finds no production importer. The three hits are:
  • a comment in extdeps.cloudflare.account_api_tokens;
  • a source string inside test.claim.text_string_importer_census_test;
  • the census lens's own fixture, test/fixture_roots/text_string_importer_census_literals/string_type_importer.dag. The census claims read that fixture as source text with the population passed as a parameter, so they never resolve its import (39/39 PASS below).
• The v2: kernel host-text String, bound only where no String declaration is visible (stacks on #12549) #12760 literal-site census confirms it by population (below).

#12760 literal-site census: summary and routing

• Instrument: v2.lens.text_string_importer_census text_string_importer_literal_sites.
• Population: the royal-lark-857 partial (gunbc#12871 comment 5940031933; batches s000–s002 and t000–t009) plus the neat-boar-16 remainder at main 841df4abf6, srv1, gunbc sha256 919e77636b97…. The remainder covers the set difference of 229 modules, artifact branch artifact/string-literal-census-841df4ab (9f34453f51).
• TSV sha256: partial ae356ebbef46777a…, remainder 43248fa4962aa168….

Population on main	Files	STRUCTURAL literal sites	Disposition
std.string_type importers	39	538	Resolved: #12512 deleted every import
v2.std.text importers or consumers	324	4812	Routed to neat-ibex-696's lane (neat-boar-16's ruling: host-text v2.std.text importers lose the import, census-classified). Spot checks are host text: embedded module sources, notes, qualified names.
src/v2/workflow/gunbc_invoke_step_emit.dag (one site)	1	1	Finding, not fixed here. A deliberate FreeMonoid<Char> position is given the literal "_", so it needs the declared Unicode-scalar unfold. Routed under gunbc.rung_drop text_boundary_identity_wall.

• Reconciliation: 5351 sites in 364 files in total. The first classification pass counted 5 files (111 sites) as importing neither text module. That was a reader artefact: all five reach v2.std.text through multi-line imports or its structural functions. So they are counted in the v2.std.text row, and gunbc_invoke_step_emit is also listed separately.
• UNRESOLVED rows, reported and not dispositioned here. They are routed to the owner of v2: kernel host-text String, bound only where no String declaration is visible (stacks on #12549) #12760's roster. 3837 of the remainder's 4427 come from src/v2/workflow/floor_grandfathered_roster.dag; they look like roster string data, spelled_other_than_text. The other reasons are callee_module_not_parsed, field_owner_not_read, positional_argument, field_not_declared, callee_module_declined, callee_not_imported_by_name and spelling_declarer_not_loaded.

Evidence (local, seed regenerated from this tree)

claim_batch --source-root dag --source-root src/v2:

• defork_type_name_census_witness_test 6/6
• self_host_symbol_identity_binding_witness_test 20/20
• text_string_importer_census_test 39/39
• text_boundary_identity_wall_witness_test 21/21
• bare_name_ambiguity_wall_witness_test 10/10

Stage0 reached a fixed point after 2 rounds.

v1 admission (gunbc.v1_maintenance_standing v1_seed_standing): the v1.compiler.coercion edit removes a roster path whose declaration no longer exists. It serves the v2 self-host by keeping the structural roster equal to the declarations that exist.

🤖 Generated with Claude Code