There is nothing to change in the way the score is coded in the text file: to represent a chord, we will only have to code all the corresponding notes with the same position.
Main changes will appear in intern model. When creating the model from text file, we will have to collect all the notes arround the current one that belong to a chord. This means that our dynamic array won't be a note-array anymore, but a chord-array. There are new structures to define, and old structures to change.
Other problems will come when polyphonic events are no chords, but long overlaps. Idea will be to make a ``score parsing'' as it was done for the score follower (see [Mat02] for more details).