Nguyễn Thái Ngọc Duy <pclouds@xxxxxxxxx> writes: > Notice that packing_data::nr_objects is uint32_t, we could only handle > maximum 4G objects and can address all of them with an uint32_t. If we > use a pointer here, we waste 4 bytes on 64 bit architecture. Some things are left unsaid or left unclear and make readers stutter a bit while reading this paragraph. We can address them with uint32_t only because we happen to have a linear array of all objects involved already, i.e. the pack->objects[] array. The readers are forced to rephrase the above in their mind ... and each of them can be identified with an uint32_t. Because we have all of these objects in pack->objects[], we can replace the "delta" field in each object entry that points at its delta base object with uint32_t index into this array to save memory (on 64-bit arch, 8-byte pointer gets shrunk to 4-byte uint). or something like that before understanding why this is a valid memory footprint optimization.
