
Changes from nauty 2.4 to nauty 2.5
-----------------------------------

* Add Traces.  The main files are traces.h and traces.c, but
  many changes were made to dreadnaut.c and nausparse.c also.

* Allow thread-safe storage if requested by configure --enable-tls
  and available.  This allows parallel running of nauty and Traces
  in separate threads.

* The makefile now creates static libraries like nauty.a in 
  addition to object files like nauty.o.

* Remove all use of types permutation, nvector, np2vector, shortish.
  These are now int.

* Add schreier.h, schreier.c and the optional use of the random
  Schreier method in nauty.  These are also used in Traces, but
  are not optional there.

* Add large-file support to all programs, assuming it is available.
  Now there is no longer a 4GB limit on files read or written on
  32-bit systems.

* Use gcc extensions like __builtin_clz() if available and not
    disabled by configure --disable-clz.
  Use FIRSTBITNZ instead of FIRSTBIT if the argument is certain
    to be nonzero.

* New macros defined in nauty.h:
   COUNTER_FMT
   PRINT_COUNTER
   SETWORDSNEEDED
   ADDONEARC
   ADDONEEDGE
   EMPTYGRAPH

* The options structure has a new boolean field schreier.

* New procedures:
  densenauty()  in naugraph.c  - simplified dense graph interface
  sparsenauty() in nausparse.c - simplified sparse graph interface
  writegroupsize() in nautil.c - write two part group size
  copy_sg() in nausparse.c     - make a copy of a sparse graph

  densenauty() and sparsenauty() are now the recommended ways to
  call nauty from a program.  See the sample programs in the package.

* Use quicksort in place of shell sort in many places.  This is 
  implemented in the file sorttemplates.c that can be used in any
  applications.

* Apply the const attribute more liberally across the code.

* The sparsegraph fields nde and *v changed type from int to size_t.
  This is to allow more than 2^31 edges on 64-bit hardware.

* sparsegraph.h and sparsegraph.c:
    Corrected definition of SG_DECL in sparsegraph.h. (The erroneous
       definition would have worked but generated warnings.)
    Added DEFAULTOPTIONS_SPARSEDIGRAPH.
    Added comparelab_tr(), testcanlab_tr(), updatecan_tr() for Traces.

* gtools.h and gtools.c:
    Now gtools.h is made from gtools-h.in by configure.
    Updated G6LEN() to work for larger graphs.
    Use large-file functions fseeko(), ftello() if possible.

* Most tools now use the random number generator in naurng.c rather
  than that in rng.c.

* gutils.h, gutil1.c and gutil2.c:
    New procedures maxcliques(), indpathcount1(), indcyclecount1(),
      indcyclecount().

* Invariants:
    Corrected getbigcells(), making a small change to invariants
       celltrips, cellquins and refinvar.

* dreadnaut:
    Sparse nauty and Traces now incorported.
    New commands: A, G, F, FF, sr, O, OO, P, PP, S, V
    w command is now in units of 2*m.
    Command-line can run commands using -o.
    M command is extended; now applies to i as well as x.
    Implement ANSI controls if requested.
    File names for > and < can be given in "..." to allow spaces.
 
* Updates to utilities:
     listg:  add -b (Bliss format), -G (GRAPE format) and
          -y/-Y (dotty format), -H (HCP format)
     labelg: add -t (Traces) and -i16 (refinvar)
     countg/pickg: add -m (vertices of min degree),
         -M (vertices of max degree), -H (induced cycles),
         -K (number of maximal independent sets)
     genrang: add -t (tree)
     genbg: add -A (antichain)
        The makefile can also make genbgL which makes larger sizes
     directg: add PROCESS feature
     shortg: -S (use sparse nauty), -t (use traces), i16 (refinvar)

* New utilities:
     ranlabg: randomly relabel graphs
     linegraphg: compute linegraphs
     subdivideg: compute subdivision graphs
     watercluster2: orient edges of graphs (by Gunnar Brinkmann)

* Version 25r2 fixed a rare bug in Traces
* Version 25r3 fixed some problems in the configure script (thanks to Daniel Grayson)

Changes from nauty 2.5 to nauty 2.6
-----------------------------------

Changes to dreadnaut:
* dreadnaut now catches control-C when nauty or Traces is running.
  This uses the global variable nauty_kill_request.
* new command "vv" to display sorted degree sequence.
* new command "r&" to relabel according to the partition.
* new command "->>" to flush the output.
* new command "B" to turn on output flushing at the end of every
  command.  Command "-B" turns it off.  Default off.
* Command with short arguments now have to be all on one line.
  Most errors cause the rest of the input line to be skipped.
* The "R" command now preserves the colouring.

Changes to nauty:
* nauty has an extra hook usercanonproc().
* The maximum number of vertices is now 2 billion.
* Many modern processors have instructions POPCNT and CLZ* that can
  help nauty.  The configuration script now looks for them and
  attempts to use them if possible.

New file formats (see formats.txt for definitions):
* sparse6 format is now extended with "incremental sparse6"
  format. Graphs in incremental sparse6 format specify only the
  differences from the previous graph.
  As yet, incremental sparse6 is only supported by copyg (which
  has new options -i/-I to write this format), listg, pickg and
  countg. For listg, pickg and countg, the -p switch might not work
  if the input is incremental.
* The new format digraph6 is for directed graphs.  There are
  procedures for reading and writing it in gtools.c.
  The following programs can handle digraph6 so far:
    labelg, shortg, ranlabg, directg, gentourng, amtog, complg,
    copyg, dretog, catg, listg, showg, converse, converseg, delptg,
    deledgeg, countg/pickg (partially), genrang (partially), genspecialg

New utilities:
*  converseg : take converse of a digraph
*  cubhamg : hamiltonian cycles in subcubic graphs
*  hamheuristic : heuristic for hamiltonian cycles
*  twohamg : partition quartic graphs into two hamiltonian cycles
*  genspecialg : generate special graphs like paths and cycles
*  gentreeg : generateg trees, based on a program of Li and Ruskey.
*  genquarticg : generate quartic graphs, written with Narjess Afzaly.
*  dretodot : reads one graph in dreadnaut format and writes a picture
     of it in dot format. You can use tools in the graphviz library
     to deal with it.
*  vcolg : colours the vertices of graphs in all distinct ways.
     If you choose the number of colours to be 2, this the same as
     adding loops in all distinct ways.
*  delptg : delete vertices.
As always, run the command with argument "-help" to get instructions.

Extra options in existing utilities:
* amtog now properly handles loops.  (Recall that loops are ok
    in sparse6 format but not in graph6 format.)
  amtog has a switch -o that allows selecting one colour class of
  a coloured graph, and -w to suppress the warning about loops.
* copyg has -z for writing digraph6 format. An undirected graph
  can be written as a digraph, but not vice-versa.
* directg default output format is directg6.
  directg has a new option -s for splitting into cases.
* dretog allows "d" command for digraph input.
* Option -B added to countg and pickg.
* complg has new option -R.
* genrang has new option -d that makes random regular graphs of
  any degree but does not guarantee uniform distribution.
  Also option -T for random tournaments.
  Some options now work for bipartite graphs; specify the number
  of vertices on each side like n1,n2.
* labelg has extra options -C and -W. These can help to determine
    what is different between two different programs that generate
    almost the same output.
* linegraphg has -t for making the total graph.
* Most utilities can take a single argument --version, which will
  print a message stating which nauty&traces package they belong to.

Other changes:
* Traces has substantial improvements.
* Extra argument macros SWDOUBLE, SWREALRANGE, SWSEQUENCE in gtools.h.
* girth() in gutil1.c got the wrong answer for K2.  Thanks to
  Sean Irvine for reporting it.
* gtools.c defines checkgline() to check if a graph input 
  line looks ok.
* The procedures in gtnauty.c, used by labelg and other utilities,
  had an old limit of 2^22 (4 million+) vertices.  This limit is
  now removed.  Also a new procedure setlabptn() was added to set
  lab/ptn according to an integer weight.
* planarg -p had a bug causing wrong output for n >= 65536.  Fixed.
* The structured type bigint has disappeared in favour of integer type
  nauty_counter, which is "unsigned long long" if that exists and
  "unsigned long" is only 32 bits, or "unsigned long" otherwise.
  This means that some plugins for geng/genbg/gentourng may need
  to be revised.
* Problems with input of graphs with 0 vertices were fixed with help
  from Kevin Ryde.  However, note that many utilities in the package
  will not work with such graphs.  It is NOT TRUE that graphs of
  order 0 are now supported.
* Stronger measures are taken to ensure that the sort order used by
  shortg is plain byte order. This corresponds to the C collation
  order, also known as POSIX, but it may be different from the
  collation order used by default on your command line. This means
  that utilities like sort, uniq, comm and join might consider the
  output of shortg to be out of order. To avoid this, define the
  environment variable LC_ALL to equal the string "C".
     bash:   export LC_ALL=C
     tcsh:   setenv LC_ALL C
  If LC_ALL is undefined, it will also be sufficient to define
  LC_COLLATE to equal "C".  The POSIX standard says that LC_ALL
  takes precedence over LC_COLLATE if both are defined, but this
  might not be true for older systems.  If you really don't want
  to change environment variables, you can compile shortg with
  -DKEEP_SORT_LOCALE but beware that some collation orders are
  are not even deterministic (i.e. different characters might
  compare equal).
