Re: [Bitcoin-development] Roadmap to getting users onto SPV clients

[Gregory Maxwell <gmaxwell@gmail.com>]

On Tue, Dec 4, 2012 at 1:57 PM, Mark Friedenbach <mark@monetize.io> wrote:
> Alan's
  :(

> UTxO meta-chain proposal becomes vastly easier to do now that
> ultraprune is merged.

No, not really. Somewhat easier due to some structural changes, but it
still needs to invent and get consensus on a normative data structure
and people need to write implementations of the required operations on
it (implementations probably required to prove performance for
consensus).  We still have to sort through the tradeoff of making a
_single_ data structure the normative merkle tree representation for
the UTxO set to the preclusion of other implementations— including
ones which are  asymptotically faster, such as a straight hash table.

There are also issues that need to be sorted out like key structure—
the most useful index for validation is txid:vout keyed, but Alan
wanted 'address' prefixed, which is not friendly for validation but
enables robust query by address— a query that the referce normal
bitcoin software doesn't even optionally support right now.  Any
disagreements on this point must be hammed out because the structure
would be normative.

> That would allow the Satoshi client to know it's
> wallet balance and operate with a >=SPV level of security during the initial
> block download, and keep them on the path of becoming a full node. If users
> can see their balances, send and receive transactions, and otherwise go
> about their business (except for mining) during the initial block download,
> would that not address your concerns?

The above said, that is all good stuff too. And I do thing starting
fast with reduced security (be it to SPV+ or SPV) is a good idea.