I read the bitcoin docs describing how btc syncs. During IBD it gets the inventories up to the tip with getblock. Why does it not skip this step and get the blocks, generating the inventories as it gets each block? Surely getting inventories is slow, unneeded and inefficient? I believe bitcoin daemon over version 0.10.0 now sync headers rather than inventories, i am talking about pre 0.10.0.
Pre 0.10 versions synced inventories as it was the way they composed a list of blocks that they then recursively request from the peer. It provides multiple benefits over just asking for all blocks since timestamp x (where is the timestamp of the last block we had).
The biggest one being that it allows you to check the peer is on the same chain as the checkpoint you are using, this means you can check the peer is on the correct chain (eg the one your checkpoints state is correct) before downloading blocks, which are considerably larger than just hashes.