Gregory Maxwell [ARCHIVE] on Nostr: đź“… Original date posted:2016-08-24 đź“ť Original message:On Tue, Aug 23, 2016 at ...
đź“… Original date posted:2016-08-24
đź“ť Original message:On Tue, Aug 23, 2016 at 8:54 PM, Kenneth Heutmaker via bitcoin-dev
<bitcoin-dev at lists.linuxfoundation.org> wrote:
> SPV is kinda broken if the wallet doesn’t do this detection. If your wallet connects only to nodes that don’t support bloom filtering, the wallet never gets updates. We have had a spike in users reporting that their wallet isn't getting updated. To compound the problem, they rescan the blockchain and lose all of their transaction history. It has caused much panic among less technical users.
>
> We believe that failing to detect the NODE_BLOOM bit is the culprit, although it is non-deterministic, so we aren't certain.
There are almost no NODE_BLOOM supporting bloom-off nodes on the
network currently. So, while supporting this is important, I am
doubtful that its the current problem you've suffered.
There are a great many fake nodes which appear to exist purely to
monitor transactions. Many do not implement enough of the protocol to
support scanning or transaction relay. (and, in fact, relaying
transactions would make monitoring less effective).
You can't count on peers on a peer to peer network to be honest and
cooperative. Implementations need to work hard to be robust to abusive
peers. Unfortunately, the design of the bloom filtering is such that
it isn't always easy (or even possible) to be robust.
đź“ť Original message:On Tue, Aug 23, 2016 at 8:54 PM, Kenneth Heutmaker via bitcoin-dev
<bitcoin-dev at lists.linuxfoundation.org> wrote:
> SPV is kinda broken if the wallet doesn’t do this detection. If your wallet connects only to nodes that don’t support bloom filtering, the wallet never gets updates. We have had a spike in users reporting that their wallet isn't getting updated. To compound the problem, they rescan the blockchain and lose all of their transaction history. It has caused much panic among less technical users.
>
> We believe that failing to detect the NODE_BLOOM bit is the culprit, although it is non-deterministic, so we aren't certain.
There are almost no NODE_BLOOM supporting bloom-off nodes on the
network currently. So, while supporting this is important, I am
doubtful that its the current problem you've suffered.
There are a great many fake nodes which appear to exist purely to
monitor transactions. Many do not implement enough of the protocol to
support scanning or transaction relay. (and, in fact, relaying
transactions would make monitoring less effective).
You can't count on peers on a peer to peer network to be honest and
cooperative. Implementations need to work hard to be robust to abusive
peers. Unfortunately, the design of the bloom filtering is such that
it isn't always easy (or even possible) to be robust.