gssw_create_score_matrix happily takes 32-bit ints in, but they aren't allowed to be above the max value for an unsigned char if you want the alignment to work. This doesn't really seem to be documented (except by realizing that the score matrix is a matrix of bytes), or checked for, so I just spent about 10 minutes wondering why vg was producing totally bogus alignments and scores when I tried to jack up various penalties into the thousands as a way to ask for, say, alignments with no mismatches.
gssw_create_score_matrixhappily takes 32-bit ints in, but they aren't allowed to be above the max value for an unsigned char if you want the alignment to work. This doesn't really seem to be documented (except by realizing that the score matrix is a matrix of bytes), or checked for, so I just spent about 10 minutes wondering why vg was producing totally bogus alignments and scores when I tried to jack up various penalties into the thousands as a way to ask for, say, alignments with no mismatches.