a LNmillisat payment HTLC has units of measure in msat and also uses Erics payment hash for all users (ABCD)
That's correct.
a commitment HTLC(different) uses only the pubkeys of the channel partners and is measured in sats
A commitment transaction with HTLC outputs uses the public keys of the channel partners, their HTLC public keys and the hash of the payment secret. See
this post again.
a commitment is an inchannel management thing that occurs after the out of channel payment of routing packets
The specs literally say that one should not forward an HTLC unless you can enforce the contract on-chain.
bolts2 (the thing your addicted to) uses examples of old old protocol where its describing examples of 'hub' payments (where channel partner is the destination)
It's exactly the opposite. Direct payments
could work just fine without HTLCs. The main purpose of HTLCs is to enable payment routing.
otherwise if alice is trying to pay Eric. if it was alice commiting to bob first. bob could then broadcast his win he never asked for.. and because its the latest commitment(in your scenario fantasy). bob cant be revoked. and so bob gets the win. eric doesnt get paid and alice is out of money.
franky1, come on. You seem to completely ignore the fact that
TWO commitment transactions are signed for each Lightning payment. The first transaction is supposed to prevent the situation you described from happening. The transaction contains an additional (HTLC) outputs with locking scripts which I described
in the other half of this post.
Bob has no real reason to broadcast his commitment transaction with HTLC outputs unless Carol claims his HTLC and Alice stops cooperating, and refuses to sign another commitment transaction without the HTLC output.
The second commitment transaction is signed once Bob sends "update_fulfill_htlc", which includes the payment preimage, to Alice. It's the transaction you have been talking about all the time.