Delete the unused draft FFI host function implementation

[robobun]

What this does

ffi/host_fns.rs was an earlier draft of generate_symbols / generate_symbol_for_function / Function::print_source_code that ffi/mod.rs still declared and re-exported, but the live implementations are the ones in ffi_body.rs, which is what every caller uses. This deletes the draft and the Function/Step/Compiled struct copies in mod.rs that only it referenced.

Split 5 of 8 from #32023, which alii asked to be broken up. Pure code motion and deletion; no intended behavior change. Stacked on claude/split/install-cli (merge order: foundations → install-cli → these, in any order among themselves).

Verification

• Compiles in isolation against the base (cargo check), and the union of all eight splits is byte-identical to Deduplicate jsc bindings, runtime api, test matchers, and shell builtins #32023's head f29ce8fe4, which CI ran at 285/286 green (the one red was the fleet-wide napi.test.ts GC-timing flake on windows x64-baseline).
• Locally on the parent head: ffi.test.js passes. The deleted file had no callers (grep for its exports resolves only to the ffi_body definitions).
• No new tests: this adds no behavior, so the existing suites above are the regression coverage.

[robobun]

^{Updated 5:05 AM PT - Aug 11th, 2026}

❌ @alii, your commit 5b3e978 has some failures in Build #91737 (All Failures)

---

🧪   To try this PR locally:

bunx bun-pr 37362

That installs a local version of the PR into your bun-37362 executable, so you can run:

bun-37362 --bun

[claude]

LGTM — pure dead-code deletion of a duplicate draft implementation.

Checked that every deleted item (Function/Step/Compiled, generate_symbols, generate_symbol_for_function, print_source_code, print_callback_source_code, Drop for Function, FFICallbackFunctionWrapper_destroy extern) has a live equivalent in ffi_body.rs, and that no code outside ffi/ referenced the removed mod.rs re-exports or types.
The deleted LIB_DIR_Z static has zero references anywhere in the tree.
The non-ffi host_fns grep hits (Response.rs, server/) are unrelated identifiers.

Extended reasoning...

Overview

This PR deletes src/runtime/ffi/host_fns.rs (469 lines) and removes ~120 lines of draft struct definitions from src/runtime/ffi/mod.rs. The deleted code was an earlier draft of the FFI Function/Step/Compiled types and the generate_symbols / print_source_code helpers. The live implementations of all of these are in ffi_body.rs, which is what mod.rs actually wires up (pub use ffi_body::FFI) and what all callers use. This is split 5/8 of #32023, broken up at maintainer request.

Security risks

None. This is a pure deletion with no new code paths, no changes to input handling, and no changes to the live FFI implementation in ffi_body.rs.

Level of scrutiny

Low-to-medium. Dead code deletion is mechanical, but FFI is a sensitive subsystem, so I verified the "dead" claim rather than trusting it:

• Grepped for every deleted public item. All generate_symbols / generate_symbol_for_function call sites are inside ffi_body.rs and resolve to the definitions at ffi_body.rs:1707/:1846, not the deleted re-export.
• Grepped for ffi::(Function|Step|Compiled|generate_symbol|LIB_DIR_Z) across src/ — no hits. Nothing outside the module referenced the draft types.
• Confirmed ffi_body.rs has its own Function (line 1890), Step (2524), Compiled (2533), Drop for Function (1920, byte-identical cleanup logic), needs_handle_scope/needs_napi_env/ffi_header/print_source_code/print_callback_source_code, and its own FFICallbackFunctionWrapper_destroy extern declaration.
• LIB_DIR_Z has zero references repo-wide.
• The removed use lines in mod.rs (c_char, c_void, NonNull, ZBox, JSGlobalObject, bun_tcc_sys as TCC) were only used by the deleted structs.
• The apparent host_fns hits in Response.rs and server/ are unrelated (_jsc_host_fns, exported_host_fns).

The two Function types coexisted without conflict because host_fns.rs imported super::Function (the draft in mod.rs), while ffi_body.rs defines and uses its own — they were genuinely parallel, and only the ffi_body one is reachable from the FFI class that JS actually sees.

Other factors

The PR is part of a maintainer-requested split of an already-reviewed larger PR, with the union verified byte-identical to the CI-green parent. The remaining mod.rs is a straightforward re-export hub with an updated doc comment. No test changes are expected for a pure deletion; ffi.test.js covers the surviving path.