currently, the feature files created by bsefextract contain unlabelled contents similar to:
# time
23.4
# spectrum
0 0 0 0
1 2 1 0
3 4 2 2
# avg-spectrum
1 2 3 4

this should be extended to contain the dimensions of the extracted features so feature files of differing lengths can still be compared, e.g. like:

 # time
time = 23.4;
# spectrum
spectrum[3,4] = {
  { 0 0 0 0 }
  { 1 2 1 0 }
  { 3 4 2 2 }
};
# avg-spectrum
avg-spectrum[4] = { 1 2 3 4 };

(feature file examples courtesy by Stefan Westerfeld)