TorchLean API

FloatLib.Floats.ExecFloat.Info.Command

Format inspection commands #

Syntax and elaboration for #float_help, #float_info, and #float_info!. Each format package registers a renderer for its configured types. The final rule reports an unsupported type when no renderer recognizes it.

Commands inspect elaborated types at compile time. Adding a format renderer requires no change to arithmetic execution.

Print a concise guide to configuring numerical types, inspecting their semantics, and finding the family-specific constructors exported by the currently imported FloatLib package.

Instances For

    Print the format, value classes, execution strategy, proof-backed operation coverage, semantic theorem groups, and documented proof limits for an executable numerical type.

    Instances For

      Focus the report on checked range, roundoff, and exactness contracts.

      Instances For

        Enable the complete theorem, backend-planning, and trust-boundary audit report.

        Instances For

          Show the detailed theorem inventory for numerical-analysis contracts only.

          Instances For

            Elaborate and fully instantiate the numerical type supplied to #float_info.

            Every family renderer uses this same command boundary before inspecting its carrier shape.

            Instances For

              Run one family renderer against a fully elaborated type without mutating the environment.

              This keeps every #float_info extension on the same command-elaboration boundary.

              Instances For

                Run a #float_info renderer only when the supplied type uses one exact ExecFloat family.

                A family mismatch raises unsupportedSyntax, which lets Lean try the next registered format renderer. The callback receives the user-facing value type, the recovered family, and the checked family arguments. Format modules supply their own report data.

                Instances For