Documentation changes for Protocol 1.6 #5
Conversation
- new method: mempool.get_info - changed method: blockchain.estimatefee Also added sections discussing the mempool transaction ordering specification.
- Updated copyright year - Updated protocol version to 1.6.0 in document - Clarified mempool ordering description - Refer to mempool ordering properly from the "status" section.
| all transactions with unconfirmed parents, and within each grouping, they are sorted by their transaction hash (as a | ||
| string). |
There was a problem hiding this comment.
note the txid is in network byte-order here, not the human-readable form
There was a problem hiding this comment.
Thanks for the review here.
In Fulcrum, internally, I keep the hashes as bytes but in human-readable byte order always (to avoid having to reverse them when presenting them to RPCs or in human-readable form) -- Electron Cash on BCH also assumes this now. This kind of throws a wrench into the works for me to have to change this... why are you keeping them in memory in non-human readable form anyway? It's just wasted cycles to do that since internally you don't really care about exact hash value you just care about communicating to the client (in human readable hex form) and to the daemon (again, in the same ordering).
Hmm... :/
Note that the only reason the hash is "reversed" when stringified in bitcoind anyway is due to a historical quirk of sorts -- the hashes (sometimes) were uint256 values treated as a little-endian integer (as is the case with a block hash). They ended up getting "reversed" so they print nicely because humans read numbers in big endian order.
But most hashes are never internally treated as integers anymore so.. the whole reversing-them-to-print is kind of nonsensical in the first place... but since that stupidity is enforced in the RPC interface and when printing hashes.. better to keep them in memory in the "ready to print" and "ready to send to RPC" form when communicating with bitcoind, clients, etc, if you ask me.
Also cleaned up some other docs and small nits.
cculianu
changed the title
See: cculianu/Fulcrum#305 and spesmilo/electrum-protocol#6