Correct, Native code depends on the underline arch. On X86 is little endian whereason PowerPC is big endian. The solution is just to make endianess correct before parsing and recostructing the IOTDB 'model'.
Another question, i would like to know more about the following abstraction in the Java code. - Datapoint (we record datapoints....tuples (key, values). - Record - Tablet There is another thing that i don't see well. It's about the index. In the docs you talk about IndexOfTimeseriesIndex but i don't see in the code such class whereas I see a lot of metadata classes. I guess that the last correct structure is: https://user-images.githubusercontent.com/33376433/123052025-f47aab80-d434-11eb-94c2-9b75429e5c54.png
