serge1
commented
4 years ago Hi,
The structures are just a "view" of data resided in section memory. And the section memory is exact to its serialized form. Converting in advance and keeping this data would require additional data layer that I would like to avoid.
The points you are raising. definitely, have the place. But, the these difficulties are on the library implementer (or researcher) and are transparent for the library user.
Best regards, Serge
First of, it's great that this library can handle 32-bit, 64-bit and little/big endian ELFs so seemlessly.
However, in terms of endian support, if the host OS is e.g. little endian and the binary being parsed is big endian, the values loaded into the structs will show up with reversed endians in debuggers which can be confusing. This is of course corrected by the library by flipping the byte order as necessary when calling one of its getters. Maybe it's better to flip every struct member's endian immediately after parsing (if necessary) so that during debugging the correct values can be seen inside the structs as well and the specific getter methods which so far handle the byte order swapping can be replaced with struct accessors instead.