I just commited a new set of sources to the SVN repository (R35) which enable reading of Guitar Pro 6 files!
But that was just the beginning. At a first step we translated the sources of TuxGuitar(Julian Casadesus) and J.Jørgen von Bargen from Java to Haxe. Since alphaTab is released under GPL like TuxGuitar we are good to go Thanks to all developers which made those algorithms for uncompressing the Guitar Pro 6 container format. All credits of this revision go to them.
As a next step I need to change how the score.gpif XML Structure gets parsed into the currently used song model which will take some time. Why? Because the new structure used by Arobas is not as hierarchical like the old one. The new format uses unique ids for tracks, bars, voices, beats, notes and even rhythms to reference inbetween the songmodel. The current algorithm uses a simplified songmodel to pre-read all elements for later accessing by the unique ids and this is what we need to change. To support all effects and elements provided by Guitar Pro 6 we would need to recreate a second songmodel which isn’t quite a good solution. How to solve this problem? I’m currently thinking of a more selective XML parsing rather than an incremental one. Simplified: I’ll try to create an algorithm which is alike the Guitar Pro 5 reading algorithm.
Stay tuned for further updates.