I perceive that SegWit is backward suitable (soft-fork) from this angle.
So, for outdated nodes the next output script (scriptPubKey)
OP_n (the place n from 0 to 16) <2-40 bytes>
… is taken into account as anyone-can-spend and in “concatenation” with the empty enter script (scriptSig) will probably be at all times legitimate.
For brand new (up to date) nodes will probably be thought-about as SegWit, so the particular SegWit validation course of, as Pieter Wuille described, will begin.
Observing solely this, every thing is evident to me.
What is just not clear to me is how all this can work when the construction of the transaction has been modified from SegWit.
Specifically, in accordance with bip144, the construction of the transaction has been modified by “increasing” the transaction with two bytes (marker (0x00) and flag (0x01)) after model discipline, and in addition including witness knowledge after tx_outs and earlier than locktime.
For brand new (up to date) nodes that is thought-about as an indication that there’s witness knowledge, nonetheless, for outdated nodes it is a transaction with 0 inputs and 1 output which is taken into account invalid so they may at all times discard this transaction (particularly since there may be additionally witness knowledge that’s not clear to them in any respect).
Even when we’d take into account that the outdated nodes wouldn’t relay these transactions to different friends, they’d not even settle for them within the blocks they obtain from the miners, as a result of for them these are fully invalid transactions (legitimate from the primary perspective of inputs and outpus since they’re anyone-can-spend, however invalid from this second, construction perspective).
Can somebody clarify to me how this works? I misunderstood one thing.