GFA BASIC Rust Documentation

Amiga GFA BASIC 3.5 — five more declaration-level audit blocks

Research Note

Amiga GFA BASIC 3.5 — five more declaration-level audit blocks

Primary source: supplied GFA BASIC 3.5 Interpreter, Chapter 13 addendum, printed pp. 13-1–13-3 and 13-34–13-36. The independently supplied combined update scan numbers the latter material Chapter 14, printed pp. 14-5–14-7. These are source-confirmed 3.5 forms; runtime conformance is not implied.

1. _DATA — query the DATA pointer

  • Context: special variable / expression value.
  • Source spelling: _DATA (leading underscore confirmed visually on printed p. 13-34).
  • Effect: reports the current DATA pointer position. The addendum says it is zero when the next READ would produce “out of data”.
  • Version: explicitly documented in the 3.5 addendum.
  • Source: printed p. 13-34; combined update p. 14-5.
  • Unspecified: numeric representation/type, whether the value is an ordinal, byte/record offset or opaque token, and behavior after program edits or source composition.

2. _DATA= — restore the DATA pointer

  • Context: special assignment form.
  • Source spelling: _DATA=value; the right-hand value is a position previously obtained from _DATA.
  • Effect: sets/restores the DATA pointer to a previously determined position, allowing subsequent READ operations to resume there.
  • Version: explicitly documented in the 3.5 addendum.
  • Source: printed p. 13-34; combined update p. 14-5.
  • Unspecified: accepted expression type, validation of arbitrary/stale values, behavior across procedures/program changes, and exact error conditions. Do not model it as ordinary assignment to a scalar variable without preserving the special pointer side effect.

3. _0 through _9 — special longword variables

  • Context: ten predefined special variables.
  • Source spellings: _0, _1, …, _9 (leading underscores confirmed on printed p. 13-35).
  • Representation: longword variables; the source compares them to integer variables with the % suffix and distinguishes them from floating-point variables.
  • Lifetime: cleared by NEW; not cleared by CLEAR, LOAD, CHAIN, or RUN, according to the addendum.
  • Version: explicitly documented in the 3.5 addendum.
  • Source: printed p. 13-35; combined update p. 14-6.
  • Unspecified: exact signedness/range wording, initial values, persistence across interpreter reset/process exit, and aliasing/interaction with memory operations.

4. MAT BASE

  • Context: matrix-operation mode statement.
  • Source forms: MAT BASE 0 and MAT BASE 1.
  • Effect: sets the row/column index offset used by the documented matrix operations. The stated default is MAT BASE 1; MAT BASE 0 resets the offset to zero.
  • Affected operations named by the source: MAT READ, MAT PRINT, MAT CPY, MAT XCPY, MAT ADD, MAT SUB, and MAT MUL.
  • Qualification: the addendum says it is sensibly used with OPTION BASE 0. The Chapter 13 matrix operations are limited to one- and/or two-dimensional fields with floating-point variables.
  • Version: explicitly documented in the 3.5 addendum.
  • Source: printed pp. 13-1–13-2.
  • Unspecified: invalid argument handling, exact interaction with arrays dimensioned under each OPTION BASE setting, and whether changing MAT BASE affects already dimensioned arrays or only operation indexing.

5. MAT CLR, MAT SET, MAT ONE — matrix/vector generation

  • Context: matrix/vector mutation statements for floating-point fields.
  • Source forms: MAT CLR a(), MAT SET a()=x, MAT ONE a().
  • Effects: MAT CLR sets every element to zero; MAT SET sets every element to scalar x; MAT ONE creates a square identity matrix with ones on the diagonal and zeros elsewhere.
  • Source equivalence: MAT CLR is described as equivalent to ARRAYFILL a(),0; MAT SET is described as equivalent to ARRAYFILL a(),x.
  • Qualification: the Chapter 13 matrix operations apply to one- and/or two-dimensional fields with floating-point variables; do not generalize to integer or string arrays.
  • Version: explicitly documented in the 3.5 addendum.
  • Source: printed pp. 13-1–13-3.
  • Unspecified: exact error behavior for non-square MAT ONE targets, zero-sized dimensions, non-floating arrays, and scalar conversion/precision.

Compatibility and evidence boundary

These five blocks are confirmed as documented Amiga interpreter 3.5 features by the version-specific addendum. Confirmation is limited to the source-described forms and behavior above. The unresolved points are intentionally not filled using another GFA dialect or generic BASIC assumptions. No Rust implementation or runtime-conformance claim is made here.