Bitcoin Forum
March 30, 2024, 02:43:04 AM *
News: Latest Bitcoin Core release: 26.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 ... 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 [76] 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 ... 200 »
  Print  
Author Topic: [SKY] Skycoin Launch Announcement  (Read 381489 times)
gomei
Hero Member
*****
Offline Offline

Activity: 770
Merit: 500



View Profile
January 21, 2015, 03:38:51 PM
 #1501

Great to be following your progress.

Please make sure you have Windows and Linux wallets as well as source before you start the IPO.
YES, we need the source before IPO.

.Ambit.    ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  █████
██
████████████
   Become part of the mining family   
✔ SECURED  │ WHITEPAPER │  ★ 171% ROI
██   
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
█████  ██
██
████████████
1711766584
Hero Member
*
Offline Offline

Posts: 1711766584

View Profile Personal Message (Offline)

Ignore
1711766584
Reply with quote  #2

1711766584
Report to moderator
The grue lurks in the darkest places of the earth. Its favorite diet is adventurers, but its insatiable appetite is tempered by its fear of light. No grue has ever been seen by the light of day, and few have survived its fearsome jaws to tell the tale.
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
gagalady
Legendary
*
Offline Offline

Activity: 938
Merit: 1000


View Profile
January 21, 2015, 05:01:47 PM
 #1502

Is there any official day for ipo.
gomei
Hero Member
*****
Offline Offline

Activity: 770
Merit: 500



View Profile
January 23, 2015, 06:00:00 AM
 #1503

The official update from skycoin is lost, wait to the new one.

.Ambit.    ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  █████
██
████████████
   Become part of the mining family   
✔ SECURED  │ WHITEPAPER │  ★ 171% ROI
██   
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
█████  ██
██
████████████
MemoryShock
Hero Member
*****
Offline Offline

Activity: 728
Merit: 500


View Profile
January 23, 2015, 06:07:16 AM
 #1504

The official update from skycoin is lost, wait to the new one.

Still looking forward to this one...he announces a launch and then goes back into dev mode inherently postponing the launch.  So completely different from the usual...-lolz-


██
██
██
██
██
██
██
██
██
██
██
██
██
██
██
  I/O DIGITAL
▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬
iodigital.io & iocoin.io

█████████████████
███████████████████
████████▌████████▐████
███████████████████████
████████████████████████
█████▌██████████████▐███
█████▌██████████████▐███
█████▌██████████████▐███
████████████████████████
███████████████████████
████████▌████████▐████
███████████████████
█████████████████
██
██
██
██
██
██
██
██
██
██
██
██
██
██
██
illodin
Hero Member
*****
Offline Offline

Activity: 966
Merit: 1003


View Profile
January 23, 2015, 06:33:09 AM
 #1505

The lunch was yesterday and the ipo lasted 24 hours, too bad bitcointalk forums went down soon after the announcement so many people missed it.
patmast3r
Hero Member
*****
Offline Offline

Activity: 980
Merit: 1001


View Profile
January 23, 2015, 07:08:38 AM
 #1506

The lunch was yesterday and the ipo lasted 24 hours, too bad bitcointalk forums went down soon after the announcement so many people missed it.

Well I had lunch anyway. What about launch though ? Tongue

xyzzyx
Sr. Member
****
Offline Offline

Activity: 490
Merit: 250


I don't really come from outer space.


View Profile
January 23, 2015, 09:03:27 AM
 #1507

OP, what is the format of the wallet files found in "~/.skycoin/wallets"?  For example, on startup my wallet file looked like this (edited out actual numbers):


{
    "id": "XXXX",
    "type": "Deterministic",
    "name": "",
    "filename": "2015_AA_BB_CCCC.wlt",
    "entries": [
        {
            "address": "Kw5Rb7XNuHqoazHpCvrb1kzpH5hcRCophj",
            "public_key": "0268ee7bddefd1d876bf12c25c1acfb182383bffab856802c30a84094ecbd7bc0f",
            "secret_key": "SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS"
        }
    ],
    "extra": {
        "seed": "XXXXYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY"
    }
}


For example, I was able to make a new wallet file by hand that didn't crash skycoin by doing the following: (Of course, when generating an address one wouldn't use such a simple and easy to guess passphrase.)

Code: (bash)
go run ./cmd/address_gen/address_gen.go -seed="foo bar baz qix"

Which outputs the public key, private key, and address of:

039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107 0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21 JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X

I assume the seed property of the wallet is a SHA256 sum of the passphrase, so I do:

Code: (bash)
echo "foo bar baz qix" | sha256sum --

Which output a seed of "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"

Placing that into the format of the above wallet file, I get:


{
    "id": "a345",
    "type": "Deterministic",
    "name": "Foo",
    "filename": "2015_01_02_3333.wlt",
    "entries": [
        {
            "address": "JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X",
            "public_key": "039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107",
            "secret_key": "0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21"
        }
    ],
    "extra": {
        "seed": "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"
    }
}


The app appears to see the wallet:



Is the above info on generating a valid wallet correct?   I have no currency so I cannot test sending to or from any of the addresses I've created, but they don't crash the app.

"An awful lot of code is being written ... in languages that aren't very good by people who don't know what they're doing." -- Barbara Liskov
cryptrol
Hero Member
*****
Offline Offline

Activity: 637
Merit: 500


View Profile
January 23, 2015, 09:50:43 AM
 #1508

For example, I was able to make a new wallet file by hand that didn't crash skycoin by doing the following: (Of course, when generating an address one wouldn't use such a simple and easy to guess passphrase.)

How do you manage to crash the app in the first place ? Can you provide the steps you follow ?
xyzzyx
Sr. Member
****
Offline Offline

Activity: 490
Merit: 250


I don't really come from outer space.


View Profile
January 23, 2015, 10:29:13 AM
 #1509

For example, I was able to make a new wallet file by hand that didn't crash skycoin by doing the following: (Of course, when generating an address one wouldn't use such a simple and easy to guess passphrase.)

How do you manage to crash the app in the first place ? Can you provide the steps you follow ?

If editing a wallet file, just randomly change any of the bytes for public or private keys, for example, will crash the app on startup.

At least it did for me.

I haven't looked at the sourcecode yet, as I don't know the go language.  It looks a lot like Oberon2, which I do know however, so I may give it a go.

"An awful lot of code is being written ... in languages that aren't very good by people who don't know what they're doing." -- Barbara Liskov
gomei
Hero Member
*****
Offline Offline

Activity: 770
Merit: 500



View Profile
January 23, 2015, 11:12:19 AM
 #1510

The lunch was yesterday and the ipo lasted 24 hours, too bad bitcointalk forums went down soon after the announcement so many people missed it.

Well I had lunch anyway. What about launch though ? Tongue
He meant the launch Grin
wait for the new announcement.

.Ambit.    ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  █████
██
████████████
   Become part of the mining family   
✔ SECURED  │ WHITEPAPER │  ★ 171% ROI
██   
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
██  ██
█████  ██
██
████████████
skycoin (OP)
Hero Member
*****
Offline Offline

Activity: 498
Merit: 500


View Profile WWW
January 23, 2015, 11:24:32 AM
Last edit: January 23, 2015, 12:05:06 PM by skycoin
 #1511

OP, what is the format of the wallet files found in "~/.skycoin/wallets"?  For example, on startup my wallet file looked like this (edited out actual numbers):


{
    "id": "XXXX",
    "type": "Deterministic",
    "name": "",
    "filename": "2015_AA_BB_CCCC.wlt",
    "entries": [
        {
            "address": "Kw5Rb7XNuHqoazHpCvrb1kzpH5hcRCophj",
            "public_key": "0268ee7bddefd1d876bf12c25c1acfb182383bffab856802c30a84094ecbd7bc0f",
            "secret_key": "SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS"
        }
    ],
    "extra": {
        "seed": "XXXXYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY"
    }
}


For example, I was able to make a new wallet file by hand that didn't crash skycoin by doing the following: (Of course, when generating an address one wouldn't use such a simple and easy to guess passphrase.)

Code: (bash)
go run ./cmd/address_gen/address_gen.go -seed="foo bar baz qix"

Which outputs the public key, private key, and address of:

039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107 0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21 JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X

I assume the seed property of the wallet is a SHA256 sum of the passphrase, so I do:

Code: (bash)
echo "foo bar baz qix" | sha256sum --

Which output a seed of "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"

Placing that into the format of the above wallet file, I get:


{
    "id": "a345",
    "type": "Deterministic",
    "name": "Foo",
    "filename": "2015_01_02_3333.wlt",
    "entries": [
        {
            "address": "JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X",
            "public_key": "039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107",
            "secret_key": "0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21"
        }
    ],
    "extra": {
        "seed": "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"
    }
}


The app appears to see the wallet:



Is the above info on generating a valid wallet correct?   I have no currency so I cannot test sending to or from any of the addresses I've created, but they don't crash the app.

Yes. Good eye. I am glad that works.

Wallet format is defined in /src/wallet/wallet.go

Wallets are in "interface". So you can implement different wallets.
- There is a deterministic wallet that implements the interface (default)
- there is a "simple wallet" which just stores private keys like Bitcoin's wallet
- If you are company you might have N wallets, one per employee and with only receiving and not sending (implement interface)
- If you are an exchange, you might want a wallet that keeps customers funds separated by customer instead of mixing (implement interface)

The problem is, the existing wallet interface is horrible. Its better than Bitcoin's but it can be improved.
- look at /src/wallet/wallet.go
- look at /src/wallet/deterministic.go

For the wallet format. I think that there should just be a map, with optional fields
- "name" : "user chosen wallet name"
- "seed" : "actually a string, but hex SHA256 works too"

Then there should be an array. The array should be
- "Address : Privatekey" or "Address Publickey Privatekey"

We do not want people giving out private key by confusing it with public key

Maybe an array of maps
{
 "address" "..."
 "seckey" : "...",
 "pubkey" : "...",
}

Also, "ID" I am not sure if ID should be in the wallet itself. Seed should be at the top of the wallet. The wallet format needs to be improved, cleaned up, simplified.

JSON API

Now that you found the wallet files, try this

go run ./cmd/skycoin/skycoin.go -web-interface=true
http://127.0.0.1:6420

Now look in
/src/gui/wallet.go

See Bottom


These functions you can call in browser
http://127.0.0.1:6420/wallets

You get an array of your wallets


I am adding functions for
- getting all unspent outputs
- there should be function for getting block depth and other information (its probably in daemon)
- getting outputs for an address
- injecting transactions

If you add enough json functions, you can build the blockchain explorer inside of the wallet. You can check balances locally, instead of having to scrape blockchain.org like people do in Bitcoin.

Adding wallet from seed:

There was one function that did three things
- create wallet
- get wallet
- update wallet

I split the functions. So they are separate. There is not a seed parameter on the wallet creation.

You can now add a deterministic wallet by seed. There will be gui button for this eventually.
http://127.0.0.1:6420/wallet/create?seed=yourseed

Overview:

Overall, the architecture is very good. However, the details need polish. This involves removing code and simplifying, removing unnecessary things. I dont feel like the Bitcoin API has changed in five years. You still cannot check balances and get outputs for an address not in your wallet, in Bitcoin. So in some sense, the Skycoin API has evolved faster than in two days than Bitcoin did since launch.

We are using JSON format, so people will not lose wallets like in Bitcoin where a new client cannot load an old wallet serialized with older version of the key-value store library. There are people who has 20,000 Bitcoin and deleted wallet because it said it has zero coins, (old wallet format, with new bitcoin client) when Bitcoind loaded wallet with silent error. You have to know there are coins in wallet and then export the privatekeys by hand, then import them by hand on command line to even know if there are coins in the wallet.

Also, JSON is good, because you can pull out a single private key or address by hand. You have to do this on command line with Bitcoin.

The seed value is good, because as long as you have that, you wont lose the coins.

Multiple wallets in client is good (will be able to switch wallets from dropdown). Most people keep a small wallet (for mobile phone) and a larger wallet. People often delete wallets with coins in them by accident, because they have to rename, move wallets in bitcoind and restart client. Skycoin will have a button from creating a wallet from seed in the web wallet gui.

There are a number of polish issues. Such as if you write a wallet to disc, but the power goes out. The write will "finish but the hard drive didnt write the data to disc and your wallet is gone. So we can't overwrite the wallet, we write a new wallet, then we flush the disc cache by hand, then we rename the first wallet and then rename written wallet to first wallet, then we write random data over the old wallet, then delete it.  You have to make sure all of those steps occur.

If you dont overwrite the old wallet with random data before deleting it, then it can be recovered from disc. If every transaction generates a new address and writes a new wallet to disc, you might have 200 copies of the wallet seed written to disc after a few hundred transactions. Someone could scan the empty space on disc and possibly steal the seed value, even if the wallet is "deleted". So you have overwrite the wallet with random data before deleting it.

I want to get it working, then get skywire prototyped and come back to polish it to that level later.

cryptrol
Hero Member
*****
Offline Offline

Activity: 637
Merit: 500


View Profile
January 23, 2015, 02:10:45 PM
 #1512

I install skycoin in Ubuntu 14.04.1 LTS, my network is available, but when I install it, it still hit the below error:

gary@ubuntu:~/Downloads/skycoin-master$ ./setup.sh
Installing gvm and go1.4
ERROR: Already installed!
Downloading Go source...
ERROR: Couldn't download Go source. Check the logs /home/gary/.gvm/logs/go-download.log
Installing skycoin to $GOPATH
GOPATH is not set
Installing or updating skycoin dependencies
xargs: go: No such file or directory
Done
Do './run.sh -h' to confirm it is installed. This runs the daemon.
README.md has further instructions for building and running the gui.

I checked the log, it says:

abort: error: Network is unreachable

So where is the link to download Go source, I want to check it in Browser to see if I can reach out it, is it https://github.com/golang/go, I can open it in my browser



GVM is a pain. Just install go from the ubuntu repo and set GOPATH to where you downloaded skycoin ( and its dependencies).
subaked
Newbie
*
Offline Offline

Activity: 3
Merit: 0


View Profile
January 23, 2015, 11:20:27 PM
 #1513

before you do the ipo
has skycoin considered dpos dac's  and what do you think of dac's dpos model for skycoin.

been watching this video on bitshares.tv called
#7 Thinking of Crowd Funding your bitcoin/crypto project? Think Again.
https://www.youtube.com/watch?v=u1fBcB0yguQ
 
and he made a good point about delegates positions being used to carry on funding project developments after the crowd sell ipo money has ran out.
dulinxu
Sr. Member
****
Offline Offline

Activity: 358
Merit: 250


View Profile
January 24, 2015, 02:56:54 AM
 #1514

OP, what is the format of the wallet files found in "~/.skycoin/wallets"?  For example, on startup my wallet file looked like this (edited out actual numbers):


{
    "id": "XXXX",
    "type": "Deterministic",
    "name": "",
    "filename": "2015_AA_BB_CCCC.wlt",
    "entries": [
        {
            "address": "Kw5Rb7XNuHqoazHpCvrb1kzpH5hcRCophj",
            "public_key": "0268ee7bddefd1d876bf12c25c1acfb182383bffab856802c30a84094ecbd7bc0f",
            "secret_key": "SSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSSS"
        }
    ],
    "extra": {
        "seed": "XXXXYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYYY"
    }
}


For example, I was able to make a new wallet file by hand that didn't crash skycoin by doing the following: (Of course, when generating an address one wouldn't use such a simple and easy to guess passphrase.)

Code: (bash)
go run ./cmd/address_gen/address_gen.go -seed="foo bar baz qix"

Which outputs the public key, private key, and address of:

039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107 0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21 JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X

I assume the seed property of the wallet is a SHA256 sum of the passphrase, so I do:

Code: (bash)
echo "foo bar baz qix" | sha256sum --

Which output a seed of "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"

Placing that into the format of the above wallet file, I get:


{
    "id": "a345",
    "type": "Deterministic",
    "name": "Foo",
    "filename": "2015_01_02_3333.wlt",
    "entries": [
        {
            "address": "JaPkYZzEiEHwhNMN1D2iLsn2FFg31KNp8X",
            "public_key": "039ed2af4adc49962682ba8777ffbc98b49286b4f319ec1ef6156814b56a82d107",
            "secret_key": "0bdf7c021fe2c6085d4fac8f81714fe5a0457e45f1815105bd7073edca694f21"
        }
    ],
    "extra": {
        "seed": "a345465605945b4223c2ce1407ea00f05695bb7ebda1b90bfed65435bfbb0bf2"
    }
}


The app appears to see the wallet:



Is the above info on generating a valid wallet correct?   I have no currency so I cannot test sending to or from any of the addresses I've created, but they don't crash the app.

Yes. Good eye. I am glad that works.

Wallet format is defined in /src/wallet/wallet.go

Wallets are in "interface". So you can implement different wallets.
- There is a deterministic wallet that implements the interface (default)
- there is a "simple wallet" which just stores private keys like Bitcoin's wallet
- If you are company you might have N wallets, one per employee and with only receiving and not sending (implement interface)
- If you are an exchange, you might want a wallet that keeps customers funds separated by customer instead of mixing (implement interface)

The problem is, the existing wallet interface is horrible. Its better than Bitcoin's but it can be improved.
- look at /src/wallet/wallet.go
- look at /src/wallet/deterministic.go

For the wallet format. I think that there should just be a map, with optional fields
- "name" : "user chosen wallet name"
- "seed" : "actually a string, but hex SHA256 works too"

Then there should be an array. The array should be
- "Address : Privatekey" or "Address Publickey Privatekey"

We do not want people giving out private key by confusing it with public key

Maybe an array of maps
{
 "address" "..."
 "seckey" : "...",
 "pubkey" : "...",
}

Also, "ID" I am not sure if ID should be in the wallet itself. Seed should be at the top of the wallet. The wallet format needs to be improved, cleaned up, simplified.

JSON API

Now that you found the wallet files, try this

go run ./cmd/skycoin/skycoin.go -web-interface=true
http://127.0.0.1:6420

Now look in
/src/gui/wallet.go

See Bottom


These functions you can call in browser
http://127.0.0.1:6420/wallets

You get an array of your wallets


I am adding functions for
- getting all unspent outputs
- there should be function for getting block depth and other information (its probably in daemon)
- getting outputs for an address
- injecting transactions

If you add enough json functions, you can build the blockchain explorer inside of the wallet. You can check balances locally, instead of having to scrape blockchain.org like people do in Bitcoin.

Adding wallet from seed:

There was one function that did three things
- create wallet
- get wallet
- update wallet

I split the functions. So they are separate. There is not a seed parameter on the wallet creation.

You can now add a deterministic wallet by seed. There will be gui button for this eventually.
http://127.0.0.1:6420/wallet/create?seed=yourseed

Overview:

Overall, the architecture is very good. However, the details need polish. This involves removing code and simplifying, removing unnecessary things. I dont feel like the Bitcoin API has changed in five years. You still cannot check balances and get outputs for an address not in your wallet, in Bitcoin. So in some sense, the Skycoin API has evolved faster than in two days than Bitcoin did since launch.

We are using JSON format, so people will not lose wallets like in Bitcoin where a new client cannot load an old wallet serialized with older version of the key-value store library. There are people who has 20,000 Bitcoin and deleted wallet because it said it has zero coins, (old wallet format, with new bitcoin client) when Bitcoind loaded wallet with silent error. You have to know there are coins in wallet and then export the privatekeys by hand, then import them by hand on command line to even know if there are coins in the wallet.

Also, JSON is good, because you can pull out a single private key or address by hand. You have to do this on command line with Bitcoin.

The seed value is good, because as long as you have that, you wont lose the coins.

Multiple wallets in client is good (will be able to switch wallets from dropdown). Most people keep a small wallet (for mobile phone) and a larger wallet. People often delete wallets with coins in them by accident, because they have to rename, move wallets in bitcoind and restart client. Skycoin will have a button from creating a wallet from seed in the web wallet gui.

There are a number of polish issues. Such as if you write a wallet to disc, but the power goes out. The write will "finish but the hard drive didnt write the data to disc and your wallet is gone. So we can't overwrite the wallet, we write a new wallet, then we flush the disc cache by hand, then we rename the first wallet and then rename written wallet to first wallet, then we write random data over the old wallet, then delete it.  You have to make sure all of those steps occur.

If you dont overwrite the old wallet with random data before deleting it, then it can be recovered from disc. If every transaction generates a new address and writes a new wallet to disc, you might have 200 copies of the wallet seed written to disc after a few hundred transactions. Someone could scan the empty space on disc and possibly steal the seed value, even if the wallet is "deleted". So you have overwrite the wallet with random data before deleting it.

I want to get it working, then get skywire prototyped and come back to polish it to that level later.


it is very hard for people who dont know about code to do......  can i send bitcoin directly to you?

TERRA - Bring profits to support community, starting from cross-border payment ❘|❘ ICO ❘|❘ DISCUSSION
skycoin (OP)
Hero Member
*****
Offline Offline

Activity: 498
Merit: 500


View Profile WWW
January 25, 2015, 08:49:41 PM
 #1515

Consensus Algorithm Update:

Whitepaper Draft: http://128.199.188.22:1337/consensus.pdf
Graphs: http://128.199.188.22:1337/



This is a histogram of convergence times. The x-axis is "rounds" the network takes to reach consensus (all nodes in the same state) for a single round consensus process.  The top algorithm is hill climbing, the middle algorithm is simulated annealing and the bottom algorithm is changing randomly between hill climbing and simulated annealing.

There is a network of nodes. Each node follows a number of other nodes. Each node has a state 0 or 1. Each node sets its state based upon randomness and the state of the nodes it is following. The process is simulated in a python discrete event simulator and includes network latency.

The network of the graph used in simulation is a real world social network, taken from the Stanford Wikipedia admin dataset.

We see that the top graph (hill climbing) converges within 10 rounds and usually in less than 5 rounds. However, there is a probability that consensus is not achieved within 45 rounds (The bar on the right on the first graph). The network reaches a local minimum and gets stuck there and never convergences, but only rarely.

One round for a network of global servers, is about 250 ms (the network latency latency between Canada and Australia).  Convergence in 5 rounds, means block times of 1 second.

For the second graph (simulated annealing), we see that there is a constant probability of convergence. This is because of "Symmetry Breaking". The simulated annealing process is driven by the "Delta" at each node. The network starts with half the nodes as 0s and half as 1s and so the delta is small. As the network becomes biased, the network converges exponentially quickly to one state or the other state. The symmetry breaking process is a random walk and can take arbitrarily long.

In the third graph (each node switches random between simulated annealing and hill climbing). The hill climbing breaks the symmetry and the simulated annealing prevents the network from getting stuck in a local minimum.

This is a graph of the commutative convergence probability function for each algorithm. Even with a real world social network graph, the probability of convergence for algorithm 3 is 1 for 20 rounds. This corresponds to 4 second blocks. The majority of blocks will achieve consensus within one second. This is competitive with credit card transactions. Bitcoin transaction confirmation times are 10 minutes.



There is an even faster one round consensus algorithm, based upon the method of 1-Replica Symmetry Breaking and the cavity method for spin glass systems in statistical mechanics. The existing algorithm is fast enough for now.

Consensus is the idea that the network reaches a ground state. There is a global energy function of all nodes. There is a local energy function for each individual node, which depends on the state of its neighbors in the graph. The network has minimum energy and is in the ground state when all nodes are in the same state (consensus). Each node has a local view of the network and acting independently achieve global consensus.

In Bitcoin the blockchain which required the most hashing to produce is the consensus chain. When consensus changes a fork becomes the longest blockchain in the tree of blockchains. A reorganization happens and transactions may be invalidated (double spending).

In Skycoin, malicious nodes can slow down or manipulate the consensus process but cannot revoke consensus of previous blocks (under most circumstances). A large enough group of influential and colluding nodes can completely control which blocks are accepted or rejected. The attacking nodes, however cannot revocate previous consensus decisions.

The consensus algorithm is based upon Ben-Ors randomized decision process over a Web of Trust (WoT) network. The nodes each node is subscribed to, represents the friends (personal contacts), peers and institutions (merchants, exchanges, websites, services) the user believes are trustworthy and reliable. Instead of spreading consensus across two or three mining pools which control 51% of the hash rate (three institutions), consensus is distributed across several dozen or several hundred of high influence nodes with network centrality which must be individually subverted or collude to influence the network.

In Skycoin, nocd trust relationships can be modified to marginalize nodes which have been shown to be involved in attacks of the network. Therefore, control of the network can be reclaimed if it falls into the control of a malicious faction. Nodes lose the influence required for a successful attack, once they have acted maliciously, as people will remove those nodes from their trust list as a result of an attack. In Bitcoin, hashing power cannot be removed from people who exploit majority control of the network.

Bitcoin has no mechanism to recover once a 51% attack occurs. Once a 51% attack occurs it will lower the price of Bitcoin, hash rate will decrease as unprofitable miners shut off equipment and the cost of subsequent attacks will be reduced. Bitcoin may suffer death spiral under a successful attack. This has not occurred with Bitcoin yet, although has occurred in several alt-coins.

The primitives in Skycoin's consensus process relay heavily on "randomization. It is impossible to guarantee consensus within a finite amount of time or guarantee the absence of blockchain forks (randomization), but the probability approaches zero exponentially quickly in time.

The final implementation will have multiple tiers of transaction and consensus mutability.
- There is a small probability of a revocation for a transaction in a 0-confirmation block.
- The probability of a revocation of a block decreases exponentially in the number of confirmations
- After N-confirmations the block cannot be revoked in the consensus process without introducing a fork (consensus diversity). The network will stop and wait rather than continuing if alternative candidates remain which have unresolved fork candidates to block depth N/2.
- Even if a fork occurs, it will not affect the majority of transactions or outputs. Most transactions and outputs will exist on both forks and there is minimal ability of an active attacker to modify the validity of transaction chains they were not involved in.
- There is a federated mutability check-point every M blocks, once the block has reached sufficient depth. This includes a hash of the full blockchain state and unspent output set. A node issuing a checkpoint should subscribe to several thousands nodes in the network (possibly all nodes) and certify global consensus. These will probably be run by developers and exchanges. An individual node should compare the output of several independent checkpoint issuing nodes. I am having trouble imaging an attack that could result retroactively introduce a fork that occurs before the last check-point. If 80% of the nodes in the network disappear because of a netsplit or nuclear war, it becomes more important to avoid issuing a checkpoint which misrepresents confidence in the network state.

Federation vs Decentralization

Skycoin's architecture relies heavily on "federalization", which is a form of decentralization but different than Bitcoin's form of decentralization, Examples:

Banking:
- centralized: a global central bank that issues money and clears all transactions through hierarchical sub-banks
- federated: each bank issues their own notes, redeemable against gold (free banking)
- federation to the user level: every user issues their own notes and credit to other users, every user is a bank (Ripple-pay (not ripple))

Communications:
- centralized: AIM/QQ, a centralized entity which forwards all communications between users
- federated: Email, a network of email servers where each server is run by a different person, each of which support multiple users as clients. users@domain.com . Users have a choice of servers and can run their own server if they do not trust any existing server.
- federation to user level: Tox, each user communicates directly with their peers

"decentralized" and "peer to peer" are vague. A protocol can be federated but have peer-to-peer replication. A protocol can both be centralized on one level, federated on another level and peer-to-peer on another level. Decentralization or federation at the user level is not always superior to decentralization either.

File Sharing:
- centralized: FTP server (which is actually federated, as a protocol, but is not federated as a source of a particular file!)
- federated: Torrents. Federation of torrent file distribution (no global centralized index of torrent files, but anyone can host an index of torrent files. Choice of multiple directories of torrent files), but peer-to-peer replication of files within an individual torrent
- peer-to-peer: Kazaa. Decentralized index, decentralized searching, decentralized downloading from peers. Notice that Kazaa is completely "decentralized" but is not as robust, useful or survivable as Torrents, because it did not have a mechanism for curation. Here, decentralization of everything was inferior to federation, because the completely decentralized service (Kazaa) was flooded with junk files, while the federated service (Torrent Trackers) was able to maintain high quality files, without creating a single point of failure which could destroy the network. Kazaa was quickly destroyed, but even destroying a single directory such as Piratebay or Demonoid has been frustrating and resource intensive. What.cd replaced its predecessor within hours of the original being taken down.

In Skycoin federation means that there are privileged servers that the client must trust (such as the mutability checkpoint servers). However, each client has a choice of which servers to trust. You can run your own server or if you are running an exchange or bank, you should be running your own checkpoint server and you should also be checking against external servers. You should only accept the consensus if there is unanimous and uncontested agreement.

This is similar to N of M signatures. You choose a list of N people and you only accept something if all N agree. With 10 servers a successful attack on 9 servers will fail as the client will detect a discrepancy. For a bank or exchange, this would mean freezing transactions until issue is resolved by a human. This represents escalation from algorithmic resolution to human judgement.

Federation enters into the design of Skycoins in several places. If you have a common protocol for exchanges and each exchange is a public key, you put a list of five public keys (exchanges) on the list of exchanges you "trust". Then you look for the best Dogecoin/Bitcoin exchange rate across all five exchanges and can execute at the best quote across the five exchanges.

In Skywire, public keys replace IP addresses. This is a cryptographic version of DNS. A public key uniquely identifies a service, server, entity or communication end-point. The protocol is immune to man-in-middle attacks and no one can impersonate a service without obtaining the private-key for the service.

Multi-round consensus

We are now working on research of the multi-round "tree consensus" algorithm. Tree consensus algorithm is the multi-block consensus process and will be the algorithm implemented for Skycoin. The tree-consensus algorithm supports consensus process concurrency (consensus debate on multiple chains concurrently) and consensus diversity (there may be one or more "consensus" chains. A  rare and severe netsplit condition or successful attack may introduce forks where the preferred consensus candidate should be resolved by hand by each individual node operator. If a transaction has been executed on all consensus chains, then it has been executed for all purposes).

Consensus diversity means that if there are five forks open for consensus and your transaction has been executed on all five forks, you can safely spend the outputs of the transaction without risk of it being invalidated and the output made retroactively unspendable on any of the forks. This is a notion of "blockchain fork forward security". Forks can be introduced in future, but not retroactively. This is enforced by all honest nodes in the network through use of local timestamping (later distributed time stamping) of block publication time and strong insistence against accepting a block into the census set which could not have existed at the block publication time.

Security of local time stamping relies on several assumptions. The most important assumption is that each node has at-least one node it is connected to, that is not a Sybil attack node. If a node is connected to 10 nodes and 9 of the nodes are Sybil attack nodes that may arbitrarily delay block propagation. as long as 1 node is not in the Sybil set then the timestamps will be reasonably accurate.

Road Map

1. paper on single round consensus
2. research on multi-round tree consensus and python simulation
3. some aspects of skywire, moving Visor on top of skywire
4. implement tree consensus in go and integrate into Skycoin
5. misc. infrastructure. Check-pointing, network monitoring, emergency alert system, merkle-DAG for parallel block downloads

#1 is in progress. #2 and #3 are happening concurrency. #5 will have to be handled by the community because we dont have time.

skycoin (OP)
Hero Member
*****
Offline Offline

Activity: 498
Merit: 500


View Profile WWW
January 25, 2015, 08:58:03 PM
 #1516

IPO Update:

We have a tox bot now. You send it your Skycoin address and it sends back a Bitcoin address. You also have to set a refund address, so that we can send coins back to you, if there is a problem.



- I am setting up system for offline generation of the bitcoin addresses on an airgapped computer, to prevent the Bitcoins from being stolen if the sever is hacked.
- There is one more bug in the Skycoin client, where wallet IDs are not working and it claims balance is zero for wallet when its not. I may end up rewriting the whole wallet module.
- The API function for adding deterministic wallet by seed, is implemented. It needs to be added to the interface.
- We need to produce windows builds.

altcoinUK
Hero Member
*****
Offline Offline

Activity: 784
Merit: 1000


View Profile
January 25, 2015, 08:59:20 PM
 #1517

I always come here and can say the same thing: it is absolutely amazing. Who is this skycoin guy, he is just amazing.
altcoinUK
Hero Member
*****
Offline Offline

Activity: 784
Merit: 1000


View Profile
January 25, 2015, 09:02:01 PM
 #1518

Based on the expertise and vision of the skycoin guy I wouldn't be surprised if SkyCoin would be a direct competitor to Bitcoin and Ethereum very soon. 
cryptrol
Hero Member
*****
Offline Offline

Activity: 637
Merit: 500


View Profile
January 25, 2015, 09:52:09 PM
 #1519

Consensus Algorithm Update:

Whitepaper Draft: http://128.199.188.22:1337/consensus.pdf
Graphs: http://128.199.188.22:1337/

404
LemonAndFries
Sr. Member
****
Offline Offline

Activity: 462
Merit: 250


View Profile
January 25, 2015, 10:49:39 PM
 #1520

IPO Update:

We have a tox bot now. You send it your Skycoin address and it sends back a Bitcoin address. You also have to set a refund address, so that we can send coins back to you, if there is a problem.

- I am setting up system for offline generation of the bitcoin addresses on an airgapped computer, to prevent the Bitcoins from being stolen if the sever is hacked.
- There is one more bug in the Skycoin client, where wallet IDs are not working and it claims balance is zero for wallet when its not. I may end up rewriting the whole wallet module.
- The API function for adding deterministic wallet by seed, is implemented. It needs to be added to the interface.
- We need to produce windows builds.


Does this mean we're getting closer to IPO launch date?
Pages: « 1 ... 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 [76] 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 ... 200 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!