instance Arbitrary ArbitraryHardPath where
    arbitrary =
        ArbitraryHardPath <$> (go =<< listOf genIndex)
      where
        go []     = return Deriv
        go (i:is) = (:| i) <$> go is

As a side note, I think the above is easier

foldl' (:|) Deriv

but of course, we still don't have mixed path with that either.

In my #200 rethink, I made this type of testing code easier by tagging path indexes as hard or soft, something like

data Bip32PathIndex = Bip32HardIndex KeyIndex | Bip32SoftIndex KeyIndex

You would then have Arbitrary instance for Bip32PathIndex, generate a random list of these, and get the hard path from that.

I think I then started using hard/soft index tags in more places in the code, and it was felt this made things unwieldy compared with current approach. I'm considering to try again but striking a more delicate balance.

instance Arbitrary ArbitraryDerivPath

is also affected.

Constants should go in its own directory, instead of Util.

Network/Haskoin/Script/Evaluator.hs:154:11: Warning:
    This binding for `encode' shadows the existing binding
      imported from `Data.Binary' at Network/Haskoin/Script/Evaluator.hs:52:21-26

Network/Haskoin/Script/Evaluator.hs:238:1: Warning:
    Defined but not used: `rejectSignature'

Network/Haskoin/Script/Evaluator.hs:289:10: Warning:
    This binding for `stack' shadows the existing binding
      defined at Network/Haskoin/Script/Evaluator.hs:94:5

Network/Haskoin/Script/Evaluator.hs:318:1: Warning:
    Defined but not used: `peekStack'

Network/Haskoin/Script/Evaluator.hs:322:5: Warning:
    This binding for `stack' shadows the existing binding
      defined at Network/Haskoin/Script/Evaluator.hs:94:5

There are also some more in the tests to fix.

The function stringToBS is implemented as Data.ByteString.Char8.pack, which truncates characters longer than 8 bits. This means that a String may have unexpected Base58 encoding.

Hey folks,

I asked this question in IRC some time ago and didn't get an answer. I'm still interested in figuring this out, in particular I'd like to mess with the mining hash function in addition to all the normal parameters. Would I be able to do that with haskoin? If so, could someone write up a brief howto?

Review the block download logic when connected to a peer that is also catching up on the chain.

I am overhauling the bip32 xpubkey/xprivkey parsing and derivation machinery

This work began when I attempted to fix #192, and I noticed the current implementation is overly complex and therefore hard to fix.

I am working on simplifying the bip32 xkey handling machinery, in branch

https://github.com/tphyahoo/haskoin/commits/rework_ExtendedKeys

Hopefully I should have something ready to pull soon.

Performance is hurt by use of the String type. I propose we move all literals we can to Text or ByteString, and also that we reduce the number of conversions between data types.

Things like Transaction inputs and outputs should go in Transaction, and not in Protocol, now that everything is in the same package.

A new transaction command, under the "transactions" section in help.

sendtimelocked acc addr amount unlockBlock [-S]

funds are spendable after unlockBlock appears in the blockchain.

Write Read instances for data types that have custom Show instances, such that read . show = id

Instead of ThreadId 84 it should just say 84.

Currently it shows TxPending when it should just say pending.

haskoin seems to compile and work fine with QC 2.7.* and stm-containers 2.5.*, it'd be great if you could lift the version bounds in the cabal file

To get correct serialized representation of address for bitcoin protocol, I had to reverse (change endianness) serialized IPv4 address, despite its encoded ByteString output is already in correct endianness. Meanwhile, to add to the confusion, prefix and port number must be left unchanged. This is seen from following function to convert from standard SockAddr:

-- make NetworkAddress from SockAddr
fromSockAddr :: SockAddr -> NetworkAddress
fromSockAddr a = case a of
    --convert to IPv6 mapped IPV4, big endian
    --https://en.bitcoin.it/wiki/Protocol_specification#Network_address
    SockAddrInet pn ha -> let 
        ips = BSL.reverse $ encode ha --WHY this must be reversed?
        Just prefix = hexToBS "0000ffff"
        lower = decode $ BSL.concat [toLazyBS prefix,ips] :: Word64
        ws = (0,lower) :: (Word64,Word64)
        in NetworkAddress 1 ws (fromIntegral pn)
    SockAddrInet6 pn fi ha sid -> let
        ws = decode $ encode ha :: (Word64,Word64)
        in NetworkAddress 1 ws (fromIntegral pn)

Maybe I'm making a mistake due to being a haskell newbie, but IMO the format (Word64,Word64) for NetworkAddress input isn't very well chosen and may cause headaches/bugs in future. I'd suggest to use (Word8, Word8....) instead.

When syncing the testnet on perra:

Nov 10 22:18:58 perra hw[3905]: Reason: Received an invalid merkle block: extractMatches: number of transactions excessively high
Nov 10 22:18:58 perra hw[3905]: @(hasko_79ftBTTz0e50fQW0XBG2LI:Network.Haskoin.Node.Peer ./Network/Haskoin/Node/Peer.hs:661:15)

Errors such as the one in https://github.com/haskoin/haskoin/blob/master/Network/Haskoin/Crypto/BigWord.hs#L312 should show the value of the offending byte in hex form.

GHC 7.10 isn't supported by LTS Haskell, so I try to use a Stackage nightly from https://www.stackage.org/nightly/cabal.config

Of course, I then fall into that altogether unpleasant place known as Cabal hell:

mlg-ell:haskoin mattleon$ cabal install
Resolving dependencies...
cabal: Could not resolve dependencies:
trying: haskoin-0.2.0 (user goal)
next goal: time (dependency of haskoin-0.2.0)
rejecting: time-1.5.0.1/installed-30e... (conflict: haskoin => time>=1.4 &&
<1.5)
rejecting: time-1.5.0.1, 1.5, 1.4.2, 1.4.1, 1.4.0.2, 1.4.0.1, 1.4, 1.3,
1.2.0.5, 1.2.0.4, 1.2.0.3, 1.2.0.2, 1.2.0.1, 1.2, 1.1.4, 1.1.3, 1.1.2.4,
1.1.2.3, 1.1.2.2, 1.1.2.1, 1.1.2.0, 1.0 (global constraint requires installed
instance)
Dependency tree exhaustively searched.

For giggles, I tried --allow-newer as well. It gets surprisingly far and then chokes:

[ 4 of 44] Compiling Network.Haskoin.Script.Types ( Network/Haskoin/Script/Types.hs, dist/build/Network/Haskoin/Script/Types.o )

Network/Haskoin/Script/Types.hs:77:10:
    No instance for (GHC.Generics.Generic PushDataType)
      arising from a use of ‘Control.DeepSeq.$gdmrnf’
    In the expression: Control.DeepSeq.$gdmrnf
    In an equation for ‘rnf’: rnf = Control.DeepSeq.$gdmrnf
    In the instance declaration for ‘NFData PushDataType’
cabal: Error: some packages failed to install:
haskoin-0.2.0 failed during the building phase. The exception was:
ExitFailure 1

Currently the byte count that an integer value can have as per DER in FieldN is set to be the value of the first octet in the stream. According to DER if the first octet's MSB is 1, the rest of the first octet encodes the number of subsequent octets encoding the length of the integer. If we are to have a maximum size of 32 bytes for FieldN, then we should make sure that the first octet is no larger. This should ensure that deserialization fails in an expected manner for garbage data.

A similar wrong interpretation of the length octet exists in signature deserialization code.

I'd like to see it integrated to haskoin.

Is any one already working on it?

In order to be able to change parameters deep inside YAML data structures in the configuration file, the YAML must merge with default instead of overwriting completely, like Yesod configuration does.

This transaction contains an output that has an invalid second public key:

https://insight.bitpay.com/tx/34040cf4494f3bc9c06befb6850a89e877c56e7bbaa77a6917820cfa6c7989d9

This output cannot be decoded by Haskoin because the first byte of one public key is not 2, 3 or 4.

Paths like m/0'/1/2' and m/0/2147483647' are not being tested properly in current test suite.

"hard follows soft" paths are legitimate paths for privkey derivations, they are even included in official bip32 test vector.

It looks like the Control.Monad.Error is being deprecated. We will have to move to Control.Monad.Except.

Network/Haskoin/Script/Evaluator.hs:31:1: Warning:
    Module `Control.Monad.Error' is deprecated:
      Use Control.Monad.Except instead

Network/Haskoin/Script/Evaluator.hs:76:10: Warning:
    In the use of type constructor or class `Error'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:76:10: Warning:
    In the use of type constructor or class `Error'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:77:5: Warning:
    In the use of `noMsg'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:78:5: Warning:
    In the use of `strMsg'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:78:28: Warning:
    In the use of `noMsg'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:122:21: Warning:
    In the use of type constructor or class `ErrorT'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

Network/Haskoin/Script/Evaluator.hs:132:43: Warning:
    In the use of `runErrorT'
    (imported from Control.Monad.Error, but defined in transformers-0.4.1.0:Control.Monad.Trans.Error):
    Deprecated: "Use Control.Monad.Trans.Except instead"

rename:

HardPath -> AllHardPath

create new new at-least-one hard segment path, and make this new type called HardPath.

per discussion in #200 .

Today the faucet's ZeroMQ API hung. The logs do not show any anomaly. I presume that the ZeroMQ API green thread stopped running.

a funny thing happend parsing the test suite from the BitcoinQt client

> runTest
Canonical Valid Script Test Cases:
  sig: [0]  pubKey: [IF 0xfd ELSE 1 ENDIF] : [Failed]
can't parse key: "IF 0xfd ELSE 1 ENDIF" error: byte decode error: demandInput: not enough bytes
  sig: [0]  pubKey: [IF 0xfe ELSE 1 ENDIF] : [Failed]
can't parse key: "IF 0xfe ELSE 1 ENDIF" error: byte decode error: Get: Invalid public key encoding

It turns out that the byte sequences 0xfd (OP_PUBKEYHASH) and 0xfd (OP_PUBKEY) inside an unexecuted IF branch are valid and ignored. But when read in an if-branch that is executed, we need to push the subsequent N bytes on the stack.

This prevents us from deserializing the byte sequence statically, meaning we can't use this instance

instance Binary Script where
    get = 
        Script <$> getScriptOps
      where
        getScriptOps = do
            empty <- isEmpty
            if empty 
                then return [] 
                else liftM2 (:) get getScriptOps

    put (Script ops) = forM_ ops put

From Network.Haskoin.Crypto.ExtendedKeys.Units.derivePrvPathVectors

, ( xprv, "m/8'", hardSubKey xprv $ XKeyHardIndex 8 )
, ( xprv, "M/8'", hardSubKey xprv $ XKeyHardIndex 8 ) 

this test passes if behavior is identical for M as for m.

I believe the convention is for M to derive a public key. It's a bit sketchy in the bip32 spec but

https://github.com/GemHQ/money-tree has this, and the current repo also has in

Network.Haskoin.Crypto.ExtendedKeys.parrsePath:

pEnd = case x of
    ""  -> Just Deriv
    "m" -> Just DerivPrv
    "M" -> Just DerivPub
    _   -> Nothing

So I think the above test needs fixing.

Right?

I found myself feeding an OutPoint in JSON form to an API based on Haskoin that I wrote, and I used the provided JSON instances. I noticed that I must provide a concatenation of the transaction ID with swapped endianness and the output index as eight hexadecimal digits. Not ideal.

cabal install --enable-tests

tests/Network/Haskoin/Script/Tests.hs:198:10:
    Not in scope: `runProgram'

tests/Network/Haskoin/Script/Tests.hs:290:33:
    Not in scope: `runProgram'

tests/Network/Haskoin/Script/Tests.hs:312:10:
    Not in scope: `runProgram'

Hello, I have cutted base58 implementation to separate package
https://bitbucket.org/s9gf4ult/base58-bytestring

My question is: do you need a patch to this package adding dependency from base58-bytestring and removing it's implementation? Or you guys maybe want to add base58-bytestring to haskoin team on github?

As suggested in the discussion #204, we could offer a utility function with type signature:

(++//) :: DerivPathI t -> DerivPathI t -> DerivPathI t

If the blockchain has been synced and the network connection is removed, Haskoin will not disconnect peers. Also, when the connection returns, peers will not be reconnected.

https://github.com/bitcoin/bips/blob/master/bip-0065.mediawiki#Motivation

I think this could be useful.

Or does this already work?

When installing haskoin after following instructions on the Stackage site to install Haskell in Ubuntu, I must issue --force-reinstalls because QuickCheck is affected by haskoin.

The version of QuickCheck present in the user’s GHC installation depends on the installation of alex and happy that is performed as the last step of Stackage instructions.

http://hydra.nixos.org/build/14260477/log/raw

I am attempting to write an implementation of the BIP-70 Payment Protocol, and would like to take advantage of some of the Haskoin types in my implementation. In the process, I've discovered that the protobuf library (which I need for serialization of the Payment Protocol messages) relies on a more recent version of deepseq than haskoin is using. Would you accept a pull request that upgrades the deepseq version?

When starting a detached hw instance, any existing log file will not be deleted, truncated, or appended to, but instead it will be overwritten. This leads to tools such as tail to show the previous contents of the log file at the end.

Hi

I am using the nightly stackage and GHC 7.10.1. To satisify the stackage nightly dependencies, a few dependency versions need to be bumped.

The relevant stackage versions are:

• deepseq-1.4.1.1
• QuickCheck-2.8.1
• either-4.4.1

I submitted a pull request.

If their is interest for this feature in Haskoin, I could get around working on a patch.
Resources on stealth payments here

When dispatching requests, the library will convert to Data.Aeson.Value early instead of using a data type for API responses. It would be good to have that strongly typed.

Use this library.

This includes—but it is not limited to—functions for:

• converting endianness
• hex conversion (use Text or ByteString instead of String)

Do not use pages, but instead use offset/count in API calls.

I would like to start hacking and contribute to this project but i can't find any specs for bitcoin protocol? can u give any pointers to any resource. I mean is there an rfc? thanx.

After syncing the faucet database from my computer to the node running the faucet, I started getting crashes at the start of hw. These crashes stopped when I synced a fresh database and removed the Bitcoin XT node from the list of nodes to connect to, letting it use its hardcoded list instead.

The Build monad is currently used for transaction validation, therefore it should go in Transaction, and not in Util.