I notice that the JavaScript SDK does not truly stream but instead loads the full FIT file into a memory buffer then loops over it, and uses DataView to deal with big and little endianess.
Why not use Node native streams and buffers with the latter also having methods built in for dealing with big and little endian values over an arbitrary number of bytes as well as encoding / extracting strings.
1. More efficient use of memory, loading approx 64kB at a time.
2. Simpler and quicker decoding.
We have a use case of validating FIT files from ultra endurance where each FIT file can easily exceed 50Mb and using the SDK it can choke the server somewhat.
