Float Str / Val

• Type: feature
• Status: done
• Owner: —
• Opened: 2026-06-06 (from todo.md §4)
• Resolved: 2026-06-18 (commit b744ab6) — float Str/Val, byte-identical to FPC, cross-bootstrap clean. Found+filed a pre-existing int→float assignment bug (feature-int-to-float-assign); ValFloat works around it with float literals.

Motivation

Str/Val are implemented for integers only (compiler/builtin/builtin.pas). Float conversion is the remaining gap. (Float intrinsics Trunc/Round/Int are already done — test/test_float_intrinsics.pas.)

Scope

• Float Val: reuse the native float parser StrToDoubleBits (lexer.inc).
• Float Str: reuse the writeln float formatter.
• Wiring is nearly free per todo.md: Val is a plain proc → add a Double overload, resolver picks by destination type. Str is parser-intercepted → desugar dispatches on the value's ASTTk (float → StrFloat, else StrInt).
• :w:d widths stay literals (matches write); expression widths out of scope (see feature-flexcolumn-directive).

If importing instead of rolling: strtod (stdlib.h, clean) — not math.h. Don't block on the C-header arc.

Acceptance

Str/Val round-trip Double values in a regression test; integer behavior unchanged.

Log

• 2026-06-06 — ticket opened from todo.md §4.
• 2026-06-18 — implemented per spec; resolved. See header.