List view

• v1.0.0

  Stabilization. No new features: docs sweep, semver commitment, API freeze.

  No due date
  0% complete0 open0 closed

• v0.9.0

  Long-lived processes. Profiling, GC fragmentation, bytecode cache: the boring speed that production demands.

  No due date

  •9/10 issues closed
  90% complete1 open9 closed

• v0.8.0

  Imports and packages. The mounted import namespace: vaults, mount tables, pkg: imports — all resolving at compile time.

  No due date

  •2/2 issues closed
  100% complete0 open2 closed

• v0.7.0

  Real embedders. Per-instance capability state and policy callbacks: two runtimes in one process stop sharing mounts and handlers.

  No due date

  •4/4 issues closed
  100% complete0 open4 closed

• v0.6.0

  Errors and edges. Diagnostics that mislead get fixed, contained perf bugs get closed. No semantic changes. Lean-up / refactor pass: the type system has grown fast (cycle, .type comparison, default clause) and several features this cycle each touched 4-5 files for what should be one concept. Goal is to shrink that surface before more features land on top of it, without changing observable behavior: - Unify TypeCheck (compiler_types.zig) and FieldTypeSpec/FieldTypeAlt (value.zig) — two overlapping type-representation systems with separate switch arms to maintain in parallel (var-decl/param checking vs struct-field/union/array-elem checking). - De-duplicate the repeated 'switch (nt.base)' dispatch scattered across vm_types.zig/compiler_decls.zig (setNamedRangeError, constructNamedType, coerceNamedTypeResult, applyNamedTypeFn, emitNamedDefault all independently enumerate int/float/decimal/string/bool/rune/array_t/map_t/enum_t). - Single canonical primitive-type-name list — Compiler.isKnownTypeName and the .type feature's isTypeNamePrimitive already disagree on what counts (any/map/array inclusion differs between them). - Attach operand-width metadata to the Op enum itself instead of disasm.zig's hand-maintained comma-separated case lists per opcode (silent wrong-disassembly is a bad failure mode for a debugging tool, not a crash). - Repo housekeeping: resolve AUDIT_FINDINGS.md, libdummy.a, libwasm_test.a sitting untracked at the repo root.

  No due date

  •15/15 issues closed
  100% complete0 open15 closed

• v0.5.0

  The language settles. Every remaining breaking semantic decision lands here: truthiness, adaptable literals, named-type completeness. After 0.5, core semantics are stable.

  No due date

  •27/27 issues closed
  100% complete0 open27 closed