• A set of internally-unused functions were removed from parseTools.js. While
  emscripten no longer needs any of these functions, there is slim chance that
  some external JS library is depending on them. Please file issues if any such
  library code is found. The removed/unused functions are:
  removePointing, pointingLevels, removeAllPointing, isVoidType,
  isStructPointerType, isArrayType, isStructType, isVectorType,
  isStructuralType getStructuralTypeParts, getStructuralTypePartBits,
  isFunctionDef, isPossiblyFunctionType, isFunctionType, getReturnType,
  splitTokenList, _IntToHex, IEEEUnHex, Compiletime.isPointerType,
  Compiletime.isStructType, Compiletime.INT_TYPES, isType.
• The example shell.html and shell_minimal.html templaces no longer override
  printErr on the module object. This means error message from emscripten and
  stderr from the application will go to the default location of console.warn
  rather than console.error. This only effects application that use the
  example shell html files.
• The version of musl libc used by emscripten was upgraded from v1.1.15 to
  v1.2.2. There could be some minor size regressions (or gains) due to changes
  in upstream musl code but we don't expect anything major. Since this is a
  fairly substantial change (at least internally) we are bumping the major
  version of Emscripten to 3. (#13006)
• Added support for specifying the text encoding to be used in response filenames
  by passing the encoding as a file suffix (e.g. "a.rsp.utf-8" or "a.rsp.cp1252").
  If not specified, the encoding is autodetected as either UTF-8 or Python
  default "locale.getpreferredencoding()". (#15406, #15292, #15426)