- make adjustment of the DebugSwitches optional.
  Enable (the old behaviour) with -debug.

- only backup/modify controlDict if it contains a DebugSwitches entry

- add -no-debug option to suppress modification entirely

- write values of processor boundaries only, for diagnostic purposes.

- parallel only, and non-legacy format only

- The dummy accessOp can be useful outside of ListListOps.
- New emptyOp for using as a filter predicate (for example).

- this protected method was previously used directly for the list
  output and had the check for 2 or more elements in it. Now simply
  test the List content and handle the output preference separately.

- generates dictionary-like content, but without a separator between
  keyword and list content.

- missing verbatim/endverbatim markers, missing "Foam::" prefix, typos

- emits the notes description only

- this class really should not be used at all anymore.
  Add a message so people recognize can find any occurrences.

- introduced a ListPolicy details to make the transition between
  a short list (space separated) and a long list (newline separated)
  more configurable.

  We suppress line breaks for commonly used types that often have
  short content: (word, wordRes, keyType).

- was WM_PROJECT_API in the environment and FOAM_API in dictionaries.

  Make these both consistently FOAM_API.
  This is a non-breaking change, since the value of WM_PROJECT_API
  (added in 1812) and/or FOAM_API is purely informative.
  For the current correct values, always use

    * foamEtcFile -show-api
    * wmakeBuildInfo -show-api

- already done in argList

- If using a non-clang compiler suite (gcc, intel, etc) the additional
  lbraries required for mesa with llvm pipelines may not be found.

  Provide a mesa_llvm configuration with in the 'vtk' config file.
  Can use the usual types of settings

     * mesa_llvm=llvm-4.0.1
     * mesa_llvm=none
     * mesa_llvm=system

- accidentally introduced by 27c62303

STYLE: trial use of brace-initialized dimensionSet

- instead of writing

      dimensionedScalar(dimensionSet(1, -2, -2, 0, 0, 0), Zero);

  we can use C++11 brace-initialization to bundle the parameters
  for the dimensionSet construction and simply write

      dimensionedScalar({1, -2, -2, 0, 0, 0}, Zero);

  Note the following is incorrect syntax (extra brackets):

      dimensionedScalar(({1, -2, -2, 0, 0, 0}), Zero);

- this allows their use as templates parameters

- use file-local function to reduce some code clutter

Feature merge OpenFOAM.org

See merge request OpenFOAM-plus!226

- reduced clutter when iterating over containers

- in 1812 propagated through the distinction between areaNormal and
  unitNormal (issue #885).

  In older versions, the normal() always meant the area-normal for
  certain of these primitive.

  However, the .org version changed this to now return the unit-normal
  instead, but with the same method name. Thus add the deprecated
  message to avoid future inadvertent uses of normal() without being
  certain which one is being meant.

- a valid() method (same as !empty() call) for consistency with other
  containers and data types

- a centre() method (same as midpoint() method) for consistency with
  other OpenFOAM geometric entities

- move constructors, construct from std::pair
- input/output of std::pair

Makes it easier when using data structures for other codes.

Had this:

    Tuple2() {}

Instead of:

    Tuple2() : f_(), s_() {}

which is the same as

    Tuple2() = default;

This meant that Tuple2<scalar, scalar>() was not being zero-initialized

- reduced clutter when iterating over containers

- a FixedList is generally small so there is little advantage
  in a compact output form for uniform content.  Eg, "2{-1}"

  By avoiding this compact form we obtain output that is also
  consistent with Tuple2, for example.

- make FixedList construct from Istream explicit

- relocate the pair_entry (HashTable) and unary_entry (HashSet) into
  the Detail namespace and add output handling.

  The output handling at this level removes the reliance on zero::null
  output (HashSet) and allows direct support of pointers.
  This means that the following now works

      HashTable<T*> tbl;
      os << tbl;

  It also means that we don't need to overload operator<< for
  HashPtrTable anymore.

- avoid delete/new when calling HashSet::set(). If the entry already
  exists there is no reason to remove it and add another one with the
  same content.

STYLE: HashTable iterators now have a val() method

- identical to the object() iterator method, but shorter to type.