[Retros] Natch 3.0 beta is out

Pascal Wassong pascal.wassong at free.fr
Fri Jul 29 14:03:12 EDT 2016


Hello,

thank you for using iNatch and Natch !

For iNatch, the version 0.7 has been created for 32 bits machines, this
is a bit outdated.  I suggest using the previous version 0.6.2, which is
a full java version, independant from 32 bits or 64 bits.  Many users
(me included) prefer the 0.6.2 version.

For Natch, you must provide the number of moves.  It is strange to want
to try Natch on a position without knowing the minimum number of moves
needed to reach it.  You could test the position while increasing the
number of moves at each try.

Best wishes,
Pascal



>>>>> "Paulo" == Paulo Peccin <ppeccin at gmail.com> writes:

  Paulo> Tried to run iNatch, but it does not start in my machine. Java8
  Paulo> installed.  The logs say:

  Paulo> !ENTRY org.eclipse.osgi 4 0 2016-07-29 14:45:04.760 !MESSAGE
  Paulo> Application error !STACK 1 java.lang.UnsatisfiedLinkError: no
  Paulo> swt-win32-3320 or swt-win32 in swt.library.path,
  Paulo> java.libary.path or the jar file

  Paulo> Also, can I use Natch to find a solution to a position, but I
  Paulo> don't know how many moves it would take to get to that
  Paulo> position?

  Paulo> Thanks, Paulo

  Paulo> On Fri, Jul 29, 2016 at 1:46 PM, Pascal Wassong
  Paulo> <pascal.wassong at free.fr>
  Paulo> wrote:

  >> Hello all,
  >> 
  >> 19 years after the first version, 7 years after the latest version,
  >> a new version of Natch is out.  It is version 3.0 beta.  You can
  >> download it from http://Natch.free.fr .
  >> 
  >> Previous version already computed the order of some moves.  This
  >> was used to eliminate position having cycles in it.  For example
  >> Pa2-a8=N-b6*Pa7 and Pa7xb6.  These move orders found for a position
  >> are now used when searching for the moves.  For example, if black
  >> must play Rh8-e8-e3-c3 and white Pe2-e4, the wPe2 won't be allowed
  >> to move before the black rook arrived at c3.  This improvement
  >> increase the speed of resolution of many problems, and makes the
  >> --watch (-k for short) preaty useless.
  >> 
  >> Many more move orders are also found.
  >> 
  >> A bug has been corrected, where some solutions where not found.
  >> 
  >> A bug has been corrected on Mac, which made the program crash when
  >> the hash-tables where full.  I was surprised to learn about this
  >> problem so late.
  >> 
  >> There are also minor changes in the output file and when using the
  >> visual option.
  >> 
  >> More details about this new version are available in the NEWS file
  >> available when you download the beta version.
  >> 
  >> There were many bugs in the numerous alpha version while developing
  >> this 3.0 version, which have been found by the testers, mainly
  >> Michel Caillaud and Thierry Le Gleuher.  Thanks to them and to the
  >> other testers.
  >> 
  >> There may still be bugs in this version.  If you find one, don't
  >> hesitate to send it to me.  The informations needed are: - the
  >> proof game, being as short as possible, that reproduces the bug; -
  >> the expected solution; - the number of the position where the bug
  >> occurs, if needed; - a clear description of what is going wrong.
  >> 
  >> The precompiled version for windows is a 64 bits version.  I could
  >> also prepare a 32 bits windows version, if enough people are
  >> interested.
  >> 
  >> Happy composing, Pascal
  >> _______________________________________________ Retros mailing list
  >> Retros at janko.at https://pairlist1.pair.net/mailman/listinfo/retros
  >> 


More information about the Retros mailing list