Instead of loading the whole disc image in memory, use the Browser's File API to fetch needed parts of the file. Has to go through some odd proxying hoops since the File object is owned by the main browser thread.
Builds but doesn't do anything interesting yet.