[Bitcoin-development] Blockchain archival

[Bitcoin-development] Blockchain archival

[Chris Evans]

[-- Attachment #1: Type: text/plain, Size: 95 bytes --]

bitcoin protocol needs an archival system so the blockchain doesn't become
too big to download

[-- Attachment #2: Type: text/html, Size: 140 bytes --]

Re: [Bitcoin-development] Blockchain archival

[rob.golding]

> bitcoin protocol needs an archival system so the blockchain doesn't
> become too big to download

Some people may want it all ...

Balance at Point-In-Time summaries (say up to the penultimate 
difficulty adjustment) would be one simple way.
And make new-adopters get up and running in minutes not days, which can 
only be a good thing.

If going that route, then solutions to the 'consolidate 
addresses/wallets' question and formal 'discard' of addresses could get 
addressed.

Rob

Re: [Bitcoin-development] Blockchain archival

[Luke-Jr]

On Saturday, September 07, 2013 11:21:31 PM rob.golding@astutium.com wrote:
> > bitcoin protocol needs an archival system so the blockchain doesn't
> > become too big to download
> 
> Some people may want it all ...
> 
> Balance at Point-In-Time summaries (say up to the penultimate
> difficulty adjustment) would be one simple way.
> And make new-adopters get up and running in minutes not days, which can
> only be a good thing.

There's no reason to require the full blockchain download before being up and 
running. Bitcoin-Qt 0.9 will (probably) have Pieter's work in this area to be 
usable very quickly, and download/verify the history in the background 
(there's no way to be completely trust-free without this).

> If going that route, then solutions to the 'consolidate addresses/wallets'
> question and formal 'discard' of addresses could get addressed.

Not sure what you mean here. Addresses and wallets are two completely 
different things. Addresses are single-use destinations that point to a wallet 
(which is itself private and unknown to the network).

Luke

Re: [Bitcoin-development] Blockchain archival

[rob.golding]

> (there's no way to be completely trust-free without this).

Not quite true, as I said balance-at-point-in-time would solve that 
(and make the storage requirements much lower)

>> If going that route, then solutions to the 'consolidate 
>> addresses/wallets'
>> question and formal 'discard' of addresses could get addressed.
> 
> Not sure what you mean here. Addresses and wallets are two completely
> different things. Addresses are single-use destinations that point to 
> a wallet
> (which is itself private and unknown to the network).

For bitcoin to grow beyond interesting experiment into global everyday 
use a number of things would have to happen, not least of which is 
taking 'average punter' into account. Whilst new ideas can filter into 
the general consciousness over time,sometimes concepts have to go with 
'what already works' :)

People's concept of money hasn't really changed in over 1,000 years - 
it remains 'something of known value i can exchange for something else'.

No-one outside of bitcoin dev's and early adopters really gets the 
one-shot concept of addresses - possibly rightly so - keeping issues of 
it lowering levels of anonymity etc out of the discussion - it doesn't 
fit with the mindset people have - it's difficult enough getting 
merchants to setup separate addresses for each client, one per 
transaction is simply a waste (of addresses, storage, blockchain size, 
numnber of inputs|outputs when spending etc)

I'm sure the wife would love a new handbag everytime she gets some 
money, but the real-world just isnt like that ;)

Addresses are perceived as the equivalent of a jar you stick your coins 
in. You can have lots of jars. Each jar can be for a specific reason or 
whatever, but the analogy is there.

Wallets are like a box you keep some of your jars in. With the added 
interesting concept that a jar can be in multiple boxes at the same 
time. Only the person with the right 'key' can open the jar and take the 
contents.

However unlike the 3 money boxes I have behind me right now - which i 
can take 1 single penny out of one and put it into another - if I want 
to move bitcoins from one addresses (jar) to another *of my own* I have 
to pay a fee. Worse still if the jar doesnt have much in it I'm denied 
that ability.

End user will neither understand why or want to pay the fee, for 
dealing with their own coins.
If a jar breaks I can just tip the contents into a new one - unless I'm 
very careless, the amount in the new one = the amount in the old one - 
people will want/need it to work like that.

Similarly if you do have all these addresses around, you may want (as 
good housekeeping) discard some of them (after moving the cash).

So having the ability to specify address to send from is essential (and 
a sadly missing feature of the QT client)

'intra-wallet' transfers with an 'also discard the sending address' 
would be a way of (once confirmed) stopping any further use of that 
address (denied any further transactions by miners ?) and when 
balance-at-point-in-time is implemented, a way of shrinking the storage 
for all other bitcoin users (who chosse not to have a full transaction 
set).


If i send luke 10, and luke sends me back 3, i have 3, luke has 7.
If luke sends me 2, and i send luke 1, i have 4 and luke has 6.
To verify my ability to send jeff 4, all that is needed is to know that 
I have 4, not all the transactions that led to that state - thats how 
its done now, thats not necessarily efficient as bitcoin grows

If luke sends me 4 more, i now have 4 again, luke has 3
If i send 1 to each of the children, they have 1 each (*4)

Having a 'family' wallet means when on holiday they can have that 
rental of quad-bikes - to send the rental company 4 the client only 
needs to know that those addresses now have 1 each in them, not all the 
previous transactions - if they didnt exist at the point-in-time 
balance, then yes, it would need to know about the luke>rob>kids 
transactions, but thats all

I moved to a new netbook recently - it took 140 *hours* to d/load and 
process the blockchain (yes the wifi was that bad), I heard from one of 
our clients that (although they only had the client running during 
working hours) that to their desktop it was over 9 days before it had 
caught up.

If all I was d/loading were the transactions since the last difficulty 
change (as one example of a fixed point), and the remaining balance on 
any not-discarded address as at that point it would have been much much 
quicker, and not be shagging my shiny new hard drive.

There's more but it's 4.45 in the morning, and I cant think coherently 
until after a few hours kip and some good coffee :)

Rob

Re: [Bitcoin-development] Blockchain archival

[Jeff Garzik]

This is all FAQ territory, and has been covered on the forums for years.

Balance-at-point-in-time is not completely trust-free, as it is a
dataset that must be bootstrapped into trust by... an earlier dataset.
 Continue this logic and you have a... chain.

There is plenty of on-going discussion on UTXO snapshotting -- UTXO
lockin for each block, or something.  This is /somewhat/ like
balance-at-point-in-time, but no one pretends it is trust-free.

The /only/ way to have a completely trust-free solution is to be able
to verify all data from genesis through $now. However, it is not
necessary for all bitcoin wallets to download and verify all those
gigabytes of data; that is what SPV mode is for.

      Jeff



On Sat, Sep 7, 2013 at 11:56 PM,  <rob.golding@astutium.com> wrote:
>> (there's no way to be completely trust-free without this).
>
> Not quite true, as I said balance-at-point-in-time would solve that
> (and make the storage requirements much lower)
>
>>> If going that route, then solutions to the 'consolidate
>>> addresses/wallets'
>>> question and formal 'discard' of addresses could get addressed.
>>
>> Not sure what you mean here. Addresses and wallets are two completely
>> different things. Addresses are single-use destinations that point to
>> a wallet
>> (which is itself private and unknown to the network).
>
> For bitcoin to grow beyond interesting experiment into global everyday
> use a number of things would have to happen, not least of which is
> taking 'average punter' into account. Whilst new ideas can filter into
> the general consciousness over time,sometimes concepts have to go with
> 'what already works' :)
>
> People's concept of money hasn't really changed in over 1,000 years -
> it remains 'something of known value i can exchange for something else'.
>
> No-one outside of bitcoin dev's and early adopters really gets the
> one-shot concept of addresses - possibly rightly so - keeping issues of
> it lowering levels of anonymity etc out of the discussion - it doesn't
> fit with the mindset people have - it's difficult enough getting
> merchants to setup separate addresses for each client, one per
> transaction is simply a waste (of addresses, storage, blockchain size,
> numnber of inputs|outputs when spending etc)
>
> I'm sure the wife would love a new handbag everytime she gets some
> money, but the real-world just isnt like that ;)
>
> Addresses are perceived as the equivalent of a jar you stick your coins
> in. You can have lots of jars. Each jar can be for a specific reason or
> whatever, but the analogy is there.
>
> Wallets are like a box you keep some of your jars in. With the added
> interesting concept that a jar can be in multiple boxes at the same
> time. Only the person with the right 'key' can open the jar and take the
> contents.
>
> However unlike the 3 money boxes I have behind me right now - which i
> can take 1 single penny out of one and put it into another - if I want
> to move bitcoins from one addresses (jar) to another *of my own* I have
> to pay a fee. Worse still if the jar doesnt have much in it I'm denied
> that ability.
>
> End user will neither understand why or want to pay the fee, for
> dealing with their own coins.
> If a jar breaks I can just tip the contents into a new one - unless I'm
> very careless, the amount in the new one = the amount in the old one -
> people will want/need it to work like that.
>
> Similarly if you do have all these addresses around, you may want (as
> good housekeeping) discard some of them (after moving the cash).
>
> So having the ability to specify address to send from is essential (and
> a sadly missing feature of the QT client)
>
> 'intra-wallet' transfers with an 'also discard the sending address'
> would be a way of (once confirmed) stopping any further use of that
> address (denied any further transactions by miners ?) and when
> balance-at-point-in-time is implemented, a way of shrinking the storage
> for all other bitcoin users (who chosse not to have a full transaction
> set).
>
>
> If i send luke 10, and luke sends me back 3, i have 3, luke has 7.
> If luke sends me 2, and i send luke 1, i have 4 and luke has 6.
> To verify my ability to send jeff 4, all that is needed is to know that
> I have 4, not all the transactions that led to that state - thats how
> its done now, thats not necessarily efficient as bitcoin grows
>
> If luke sends me 4 more, i now have 4 again, luke has 3
> If i send 1 to each of the children, they have 1 each (*4)
>
> Having a 'family' wallet means when on holiday they can have that
> rental of quad-bikes - to send the rental company 4 the client only
> needs to know that those addresses now have 1 each in them, not all the
> previous transactions - if they didnt exist at the point-in-time
> balance, then yes, it would need to know about the luke>rob>kids
> transactions, but thats all
>
> I moved to a new netbook recently - it took 140 *hours* to d/load and
> process the blockchain (yes the wifi was that bad), I heard from one of
> our clients that (although they only had the client running during
> working hours) that to their desktop it was over 9 days before it had
> caught up.
>
> If all I was d/loading were the transactions since the last difficulty
> change (as one example of a fixed point), and the remaining balance on
> any not-discarded address as at that point it would have been much much
> quicker, and not be shagging my shiny new hard drive.
>
> There's more but it's 4.45 in the morning, and I cant think coherently
> until after a few hours kip and some good coffee :)
>
> Rob
>
> ------------------------------------------------------------------------------
> Learn the latest--Visual Studio 2012, SharePoint 2013, SQL 2012, more!
> Discover the easy way to master current and previous Microsoft technologies
> and advance your career. Get an incredible 1,500+ hours of step-by-step
> tutorial videos with LearnDevNow. Subscribe today and save!
> http://pubads.g.doubleclick.net/gampad/clk?id=58041391&iu=/4140/ostg.clktrk
> _______________________________________________
> Bitcoin-development mailing list
> Bitcoin-development@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/bitcoin-development



-- 
Jeff Garzik
Senior Software Engineer and open source evangelist
BitPay, Inc.      https://bitpay.com/