{fv|fa}Mesh classes inherited the [old] data class (renamed meshState)
- meshState removed from inheritance list and added as data members

ENH: added 'mesh' dictionary to meshState to hold mesh properties

- solver information now stored in a 'solver' dictionary (was solverPerformance)
  - {first|final}Iteration entry now stored in solver dict instead of top level
- mesh data (new) stored in 'mesh' dictionary

- only affects recent develop
- fixed by supplying invalid casename for non-master ranks

ENH: cyclicACMI: add non-blocking matrix updates.

See merge request !641

ENH: Reduced-order modelling field reconstruction with DMD

See merge request !639

- filehandler and watched files changes
- addition of turbulence submodule

- ensure that operator<< and operator>> behave symmetrically

STYLE: use globalIndex::totalSize()

SUBMODULE: updated cfmesh

improve fileHandler support for redistributePar

See merge request !605

mattijs authored 1 year ago and Andrew Heather committed 1 year ago

- uses read&write handlers
- fixes initialisation of partial handlers
- 'uniform' copied to wrong place
- -overwrite not deleting old processors dirs

mattijs authored 1 year ago and Andrew Heather committed 1 year ago

BUG: collated: cooperate with non-parallel

mattijs authored 2 years ago and Andrew Heather committed 1 year ago

- use file handlers to manage where meshes/fields are to be read and
  written.

mattijs authored 2 years ago and Andrew Heather committed 1 year ago

- code is compiled dynamically on the master node.
  In the normal (non-distributed) case, simply poll the NFS
  to see when it appears on the sub-procs.

  For a case with distributed roots, first broadcast it (via MPI)
  to the IO master nodes and then poll afterwards.

- on startup, detect and create missing processorXXX/ subdirectories
  on distributed filesystems

mattijs authored 2 years ago and Andrew Heather committed 1 year ago

- this delay the communication of file watches, which helps avoid
  communication deadlocks with master-only reading.

Co-authored-by: Mark Olesen <>

mattijs authored 1 year ago and Andrew Heather committed 1 year ago

- when reading, detect all clouds on all processors and uses this when
  reading fields. Similarly, when writing it uses writeOnProc to skip
  clouds that are empty on any particular processor.

Co-authored-by: Mark Olesen <>

Extend some Pstream, PstreamBuffers and globalIndex functionality

See merge request !640

- was previously limited to 'char' whereas gatherv/scatterv
  already supported various integer and float types

STYLE: rebundle allToAll declarations with macros

ENH: provide a version of allToAllConsensus returning the Map

- simplifies use and avoids ambiguities in the send/recv parameters

- the Map version will now also transmit zero value data if they exist
  in the Map. Unlike the List version, zero values are not necessary to
  signal connectivity with a Map.

COMP: forwarding template parameters for NBX routines

ENH: consolidate PstreamBuffers size exchange options

- had a variety of nearly identical backends for all-to-all,
  gather/scatter. Now combined internally with a dispatch enumeration
  which provides better control over which size exchange algorithm
  is used.

DEFEATURE: remove experimental full-NBX PstreamBuffers variant

- no advantages seen compared to the hybrid NBX/PEX approach.
  Removal reduces some code cruft.

DEFEATURE: remove experimental "double non-blocking" NBX version

- the idea was to avoid blocking receives for very large data transfers,
  but that is usually better accomplished with a hybrid NBX/PEX approach
  like PstreamBuffers allows

- these help when building upper or lower connected topologies.
  The new findProc() method is a non-failing whichProcID alternative

- provide a globalIndex::calcOffsets() taking an indirect list, which
  enables convenient offsets calculation from a variety of inputs.

- new CompactListList unpack variant: copy_unpack()
  The copy_unpack() works somewhat like std::copy() in that it writes
  the generated sublists to iterator positions, which makes this
  type of code possible:

      CompactListList<label> compact = ...;
      DynamicList<face> extracted;

      compact.copy_unpack<face>
      (
          std::back_inserter(extracted),
          labelRange(4, 10)
      );

  -and-

      const label nOldFaces = allFaces.size();
      allFaces.resize(allFaces + nNewFaces);

      auto iter = allFaces.begin(nOldFaces);

      iter = compact.copy_unpack<face>(iter, /* selection 1 */);
      ...
      iter = compact.copy_unpack<face>(iter, /* selection 2 */);

ENH: globalIndex resize()

- can be used to shrink or grow the offsets table.
  Any extension of the offsets table corresponds to '...

- report location with previous good offset and the new count that
  would cause overflow. Simpler to report and the (very long) list
  of input sizes is not particularly useful for diagnostics either.

ENH: add globalIndex comparison operators

- for outputting lists of globalIndex

- allows construction of string tokens holding character content.
  For example, data that has been serialized and buffered and that
  now needs to be written or sent to another process.