Clarification: saving offset in binary

[jozefhopko]

First of all, great project and implementation!

I was just wondering what is the motivation of using offset in binary for struct, string, vector etc. As described in the specification, these types are written by using [offset] which points to --> [size] [data..]

What is the advantage of using the offset for these data types? Why not just to write it directly after each other like bool or array attributes?

Thanks in advance!

[chronoxor]

Consider this example

struct
{
  bytes field1;
  string field2 = "";
  byte[] field3;
}

in the struct definition you'll have 3 offsets of unit32 pointing to the corresponding data later in buffer

[uint32(field1-offset)],[uint32(field2-offset)],[uint32(field3-offset)],[field1-size,field1-data],[field2-size,field2-data],[field3-size,field3-item1,...,field3-itemN]

[jozefhopko]

Great, thanks for the fast response!

Using your example, I was wondering, why not just remove the [uint32(field1-offset)],[uint32(field2-offset)],[uint32(field3-offset)]

and write directly [field1-size,field1-data],[field2-size,field2-data],[field3-size,field3-item1,...,field3-itemN]

This would reduce the size and parsing complexity. Or am I missing a use-case where the offset has an advantage?

[chronoxor]

You need to know offsets to access struct fileds, otherwise you need to travel from filed1 to fieldN. Without offsets your data will be variadic size.

[jozefhopko]

Perfect, makes sense, thanks for the clarification!