I was troubleshooting why my own program wouldn't handle alternate endings correctly in some cases and support for another Guitar Pro based application called Go PlayAlong pointed out to me that my program's import logic (which is largely based off of TuxGuitar's) was reading the alternate ending number too early in the file, causing the wrong value to be read in some circumstances like when the measure also contains a time signature change. I checked the current revision of the GP5 import code here on Sourceforge and it looks like the error is still present. The correct order in which the measure data should be read is:
Number of repeats
Key signature change
"Beam eight notes by" values
Number of alt. ending (treated as padding if the measure is not an alt. ending)
1 byte of padding
Basically the current source is reading the alt. ending number before the key signature change definition, leading to the wrong values being used.
Log in to post a comment.