Old News

Version 0.4.1 has been released. This version is mainly a bug-fix release. It fixes the following problem:

• The program could not assembler trivial problems with the new rhombic grid. This could result in some solutions not found when your problem contains very simple pieces
• The 255 piece limitation is now only applied when the disassembler is used. The assembler can handle more pieces. It is just the disassembler that is right now limited to 255 pieces.

Version 0.3.1 has been released. This version is mainly a bugfix release. It fixes the following problems:

• The assembler sometimes didn't find existing solutions. This happened when the total of all voxels defined by lower ranges was larger than the number of normal voxels in the result
• The solver did abort in cases when piece ranges were used and some pieces didn't have a mirror partner piece
• A Possible crash with very larger grids
• A missing symmetry in the triangle grid
• A crash in the sphere grid when empty (no used spheres) pieces were used
• A crash when animating a solution while the solver is running and that solution has already been dropped due to a limit on the number of solutions

There are also a lot of improvements to the built system. BurrTools can now properly crosscompile, includes its NSIS-script, doesn't build unneeded table generators, contains better checks for OpenGL. Volker Grabsch did most of that work.

Joe Becker edited the whole manual and fixed a lot of Ronalds and mine Germanizmen within the document.

The new assembly method was improved so that is is quite a bit faster in some (rare) situations

Finally the program also contains a new feature. It is now possible to change pieces within the 3D-view. To add a voxel you need to click onto the face of a voxel while pressing the SHIFT or ALT key. Then the voxel touching that face will be added. Removing voxels it possible by clicking on the voxel while pressing CTRL. Additionally selection of shapes is possible in the problem view. And hiding of pieces can be done in the solution view by clicking onto the piece while pressing SHIFT.

Version 0.2.3 has been released. This version fixes a bug that was sent in by R. J. Grimes. I ocured in puzzles that contain pairs of mirror pieces where one piece is not oriented by simply mirroring one piece along the x-axis. I never encountered this bug because I always entered the shapes by using the mirror transformation.

The result was again that too many solutions were removed.

Also from this version on I will no longer distribute the Linux binary. The chances that it works are way too small, BurrTools became simpler to compile because all libraries are now distributed in many Unix distributions. Finally BurrTools appeared in the first distribution: FreeBSD and hopefully more will follow.

Version 0.2.1 has been released. This version improves the following things:

• Many new symmetry groups, many thanks to Markus Götz
• add a red - blue/green/cyan 3D mode (anagylph mode)
• fix some bugs with sphere puzzles:
  • placement browser didn't work properly
  • shapes are now centered and don't jump around when adding new spheres
  • some puzzles where unsolvable due to internal shortcomings
• add a progress display for the creation of the status window as calculating the information in the status window might take quite some time, some other smaller improvements
• the zoom slider zooms much larger
• huge speedup in preparation time. The program used to do a lot of unnecessary checks at the start. Now only the required ones are left. This makes 10% speed saving in my regression and speed test suite. (something like 80% savings for the sphere puzzles as the preparation has been especially wasteful there)
• improvements to build system to install the documentation and leave out the headers for the library, include FLU and generally make compilation a bit easier and correct for those compiling on their own.
• BurrTools now adds a default rule with each added constraint colour and upon creation of the problem it adds rules for all defined colours
• fix a problem, when problems contain empty shapes
• the check for mirrored solutions was not always enabled when it was required resulting in additional solutions
• the mirror check was not able to correctly handle 2 pieces of identical shape. They should not be there but when they are they are now handled right

Version 0.1.9 of the BurrTools is now available. The changes in this version include

• Countless changes to the interface. Many new functions to make editing of puzzles easier and faster and to make the whole program more userfriendly. This comes with Icons made by Ronald Kint-Bruynseels
  • Comments for puzzles
  • Names for pieces, problems without names (just a number)
  • Exchanging of pieces, shapes and problems (without destroying found solutions)
  • Proper cursor in 3D-View
  • Symmetric edit modes (like in PuzzleSolver3D)
  • More ergonomic interface
  • ...
• New documentation. Ronald Kint-Bruynseels has rewritten the first part of the documentation turning it into a very nice document that contains much more information than the old one with many examples, hints of how to do this and that and lots of other things. I suggest you at least skim through it once.

  The quality of the document has also dramatically improved, the images look better, the fonts are now vector fonts and look much smoother, even with older acrobat reader versions. And all of that fits into a file that is only slightly bigger than the old documentation

• Speed improvements for many puzzles
  • with many identical pieces. The time to solve JayCube on my computer dropped from 2.5 hours to 7 minutes.
  • with many placements for their pieces. The preparation phase has been speed up considerably. The solving time for Nancy Alliegos IPP 25 exchange puzzle dropped from more than 6 Minutes to 15 seconds
  • where many cubes are always occupied by the same puzzle piece, meaning they are either all empty or all occupied by piece a or all occupied by piece b or so. This happens, for example when the basic unit of the puzzle is not a cube but you emulate it with cubes, e.g. Again Nancys puzzle. Here the basic unit is a 3x3x3 cube with a hole through the middle.
  • all calculation intensive puzzles due to hand optimized mashine level code. This is only available for intel PCs sorry. The changes result in about 20% less time to search the possibilities
  • All in all I was able to halve the time my set of test puzzles requires to run through completely.
• Creation of images with single shapes, a whole problem or solutions
• Pieces are no longer minimized when the puzzle is solved
• Many bugfixes including:
  • Problem in rotation removal, sometimes the program found rotations after all
  • With mirrors it was the other way around. It could happen that real solutions were removed. Now it can happen that some mirrored solutions are left in the list of solutions
  • Memory leak that filled up memory when there were many undisassembable assemblies
  • Crash when removing last color is fixed
  • Always ask when quitting the program, even when pressing the x in the window corner
  • Don't try to disassemble 1 piece assemblies
• Windows installer that is better suited for this small project. The new installer doesn't quatruple the size of the file to download.

Version 0.1.8a is now available. This version has accumulated important changes, here are some of the more interesting:

• BurrTools wont find rotated and mirrored solutions any longer. This has been the reason for the long time that this update took. But now it should be a quite general solution
• When BurrTools crash they should display an error message and also they should try to save the current puzzle into a file
• Usability enhancements (Key-Shortcuts, better variable cubes, larger edit grid, ...)
• Documentation updates. The included PDF now contains some more information about the internals of the program and explains some of the used algorithms. Thanks to Dic Sonneveld, Ronald Kint-Bruynseels and Mineyuki Uyematsu I've been able to add a few real puzzles that demonstrate the important features of BurrTools.
• Besides the windows binary Peter Knoppers has provided a windows installer. But because this file is more than 4 times as big as the simple compressed archive I continue providing the zip file for those with a tight connection to the net.

For a complete list look at the NEWS file inside the distribution.

Shortly before the release a problem was found. Sometimes BurrTools requires a very long time to solve simple problems. A puzzle was found that takes 500 times as long as PuzzleSolver 3D needs. As I was unable to find the solution quickly, didn't want to delay the release any longer and considered that the problem has always been there I release with this known problem.

Version 0.1.7 released. This release features quite some interesting updates:

• last version missed the placement optimisations for pieces resulting in an eternal solving time for some more complex puzzles like Bill Cutlers SM or Lovley18. This version switches the optimisations back on, that makes some puzzles a bit slower but on the other hand the solving time for some complex puzzles drops down considerable (from years to hours) Downside of this is that you can NOT continue stopped puzzles that were saved with version 0.1.6
• huge speedup for the disassembler for puzzles that have lots of assemblies for EasyLiving for example only 25% of the time is left (time dropped from 24 to 6 seconds) and also for complex puzzles like Pelikan Burr time dropped to 66% (from 5.5 to 3.6 seconds). Downside of this is that old disassemblies don't work any more. You have to solve them again.
• new editing possibilities: drag to fill rectangles and a button that makes your changes not just to the current layer but to all
• add possibility to load PuzzleSolver3D files (this can be done with the "Import" entry in the main menu)
• add function to make it possible to not completely disassemble the puzzles. This is done by putting pieces into groups. All pieces that are in the same group can stay together and the puzzle is still considered as disassembled. Group 0 is different. All pieces in group 0 need to be taken apart until single. You can set the group for each piece with the G+ and G- buttons in the problem tab
• when puzzles where saved bevore a result shape was defined the program refused to load the file. This works now. Old files can be loaded as well