Quote from: hgmichna on February 13, 2014, 08:41:16 AM

Quote from: un_ordinateur on February 12, 2014, 09:20:43 PM

Adding any new requirement would "break" the protocol. Changing the behavior of signature production/requiring a new field in the transaction indicating which input signed the txid, would make all old wallet/miners/clients/exchange software unoperative.

It would require forking the chain, because the blockchain does not actually contain that info. (I guess though the whole transaction signature might be dropped when the transaction is included in the chain; but then it would be only an "voluntary" rule enforced by miner to NOT include a mutated transaction by a third party.)

Anyway, such "breaking" behavior is generally undesirable.[]

There is always the possibility to create an alternative format and make it mandatory, starting with block 12345 some time in the future.

Obviously the clients would have to be able to process both the old format (before block 12345) and the new (block 12345 and beyond).

It is a fork, but it could be pre-announced well in time, if a large majority of miners agree.

Of course! Bitcoin has already done that a few time.

But even if the change makes total sense, and all miners agree to it, it is still something the devs try to avoid as much as possible. Because it renders all unupdated clients non-functionnal. One could say "they just had to update", but sometimes it is not easy. For example, I remember seeing a pool table which could accept bitoin payments to release the balls. It is embedded electronics. Updating that is not easy. In any case, it is bad PR to have some people's software break because of an update, even it is their own fault.

Thus the devs usually try to bunch such "breaking" changes in a single update to minimize the need for such updates, and only do it when a compelling problem requires it. The mutability problem had not been compelling enough before, but it is now.