[Dev] Segregated witness and BIP 102

Wellenreiter

@aciddude said:

@wrapper said:

I agree that it isn’t high priority, like getting the 09.3.1 version compiling on Ubuntu LTS, is much more urgent.

Boom! I love a challenge - Will create a VM when I get home and do this. I’ll aim to write the same style tutorial i did for Centos 7 Minimal.

just can you confirm it’s Ubuntu 16.04 LTS ?

D
I’m using opensuse build service, so I don’t need to set up a VM for each distribution when I build the packages.

If you are interested in the build service send me a PM.

Currently I get this error messages:

CXX      rpcrawtransaction.o
[  590s] In file included from /usr/include/boost/move/detail/type_traits.hpp:34:0,
[  590s]                  from /usr/include/boost/move/core.hpp:54,
[  590s]                  from /usr/include/boost/move/utility_core.hpp:29,
[  590s]                  from /usr/include/boost/move/utility.hpp:28,
[  590s]                  from /usr/include/boost/thread/detail/move.hpp:27,
[  590s]                  from /usr/include/boost/thread/lock_types.hpp:11,
[  590s]                  from /usr/include/boost/thread/pthread/mutex.hpp:16,
[  590s]                  from /usr/include/boost/thread/mutex.hpp:16,
[  590s]                  from allocators.h:13,
[  590s]                  from serialize.h:9,
[  590s]                  from bignum.h:9,
[  590s]                  from chainparams.h:9,
[  590s]                  from base58.h:17,
[  590s]                  from rpcrawtransaction.cpp:6:
[  590s] /usr/include/boost/variant/get.hpp: In instantiation of 'typename boost::add_reference<T>::type boost::strict_get(boost::variant<T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19>&) [with U = const CScriptID&; T0 = CNoDestination; T1 = CKeyID; T2 = CScriptID; T3 = CStealthAddress; T4 = boost::detail::variant::void_; T5 = boost::detail::variant::void_; T6 = boost::detail::variant::void_; T7 = boost::detail::variant::void_; T8 = boost::detail::variant::void_; T9 = boost::detail::variant::void_; T10 = boost::detail::variant::void_; T11 = boost::detail::variant::void_; T12 = boost::detail::variant::void_; T13 = boost::detail::variant::void_; T14 = boost::detail::variant::void_; T15 = boost::detail::variant::void_; T16 = boost::detail::variant::void_; T17 = boost::detail::variant::void_; T18 = boost::detail::variant::void_; T19 = boost::detail::variant::void_; typename boost::add_reference<T>::type = const CScriptID&]':
[  590s] /usr/include/boost/variant/get.hpp:284:25:   required from 'typename boost::add_reference<T>::type boost::get(boost::variant<T0, T1, T2, T3, T4, T5, T6, T7, T8, T9, T10, T11, T12, T13, T14, T15, T16, T17, T18, T19>&) [with U = const CScriptID&; T0 = CNoDestination; T1 = CKeyID; T2 = CScriptID; T3 = CStealthAddress; T4 = boost::detail::variant::void_; T5 = boost::detail::variant::void_; T6 = boost::detail::variant::void_; T7 = boost::detail::variant::void_; T8 = boost::detail::variant::void_; T9 = boost::detail::variant::void_; T10 = boost::detail::variant::void_; T11 = boost::detail::variant::void_; T12 = boost::detail::variant::void_; T13 = boost::detail::variant::void_; T14 = boost::detail::variant::void_; T15 = boost::detail::variant::void_; T16 = boost::detail::variant::void_; T17 = boost::detail::variant::void_; T18 = boost::detail::variant::void_; T19 = boost::detail::variant::void_; typename boost::add_reference<T>::type = const CScriptID&]'
[  590s] rpcrawtransaction.cpp:299:77:   required from here
[  590s] /usr/include/boost/variant/get.hpp:178:5: error: invalid application of 'sizeof' to incomplete type 'boost::STATIC_ASSERTION_FAILURE<false>'
[  590s]      BOOST_STATIC_ASSERT_MSG(
[  590s]      ^

The same code is compiling fine on Ubuntu 15.04, Debian 8 and other distributions.

wrapper

Just reading about a similar problem compiling :

I think that the problem is that you put #ifdef instead of #ifndef at the top of your header.h file

a search for what includes boost showed - checkpoints.h

Which is a likely area for an error as it is included code for FTC. However, I can’t see a logical error in variable definition there, yet …

https://github.com/FeatherCoin/Feathercoin/blob/0.9.3.1/src/checkpoints.h

Also note https://github.com/FeatherCoin/Feathercoin/blob/0.9.3.1/src/rpcprotocol.cpp

last change was to fix compile error …

Also noted libqt5core5a replaces libqt5core5 mentioned in master-0.8/doc/readme-qt.rst

lizhi

@wrapper

Segregated Witness is a great improvement. It is not in order to expand the blockchain size.

Because tx_hash = double_sha256(raw_tx), When S’ value replace S value, a unique transaction may have 2 different hash (tx_hash). It is a nightmare scenario.

The transaction is the transaction, the signature is the signature. So we need to take out the signature, put it out, and put it on the outside. signature insert into witness

Transaction structure = TxIn + TxOut.

Currently a total of 4 related BiP (141-144) is proposed.
https://github.com/bitcoin/bips/blob/master/bip-0141.mediawiki
https://github.com/bitcoin/bips/blob/master/bip-0142.mediawiki
https://github.com/bitcoin/bips/blob/master/bip-0143.mediawiki
https://github.com/bitcoin/bips/blob/master/bip-0144.mediawiki

wrapper

@lizhi said:

@wrapper

Segregated Witness is a great improvement. It is not in order to expand the blockchain size.

Cheers for that and all your great work, I am for further developments in the software. I will study up as soon as I get a chance.

bsotnikow

@wrapper Is segwit on the road to smart contracts? From what I understand they need to implement segwit on BTC before they’ll be able to do smart contracts with it… . .I also for a long time didn’t see the value in smart contracts, but now I see. You can basically set up escrows right? Wherein coins can be tied up until certain requirements are met? That is a massive addition of function AND a very marketable thing towards the general public.

wrapper

'Lizi & @Bostnickow – Actually that sounds great

I will read up on it more urgently, (am still failing to compile 9.3.1 on Ubuntu 15.10)

but our basic policy is to include such in a “Development” version, then try to get some enthusiasm to test it, we have set up test networks for specific features, it seems to me testing a smart contract feature may attract more attention…

Some of these features are complex, or really need to be on the network to be fully tested, at worst case that need high supervision at best they are backwardly compatible.

lizhi

0.9.3.2 is a bridge.It is compatible with the new blockchain version. Please upgrade client.

bsotnikow

Events with ETH have changed my mind on this topic. I don’t think its a good idea to add complications like Segregated witness to the code. Keep the code as simple and secure as possible, focus on being a savings instrument and profitable for miners… It may take longer, but eventually someone will figure out how to create an encrypted bit of software that can create and manage crypto wallets. You’ll probably be able to download them free one day and set up as many smart contracts as you like with whatever crypto you like.

ghostlander

@lizhi Your patch allows to put ANY block version above 2 in new blocks. Like 102938. It defeats the purpose of block versions at all. If someone puts maliciously such a block version in a new block, we have a forked network between all existing wallets and v0.9.3.2.

https://github.com/FeatherCoin/Feathercoin/commit/911098d4b9124ff01406729efc345a4dbbff6d68

Don’t upgrade to v0.9.3.2 unless you’re ready for a trouble.

wrapper

We’re going to test some interface changes and do some help for the new features in 0.9.3.2.

It would be possible to make a intermediate release “0.9.3.3” with just the UI and minimum - fixes –

Then I’ll move onto into 0.1x.x series, and transfer those in and seeing if I can add some help and sharpen the new forms further. (any assistence accepted)

I’m assuming we can still make that branch compatible, while we create a tidy release of that.

https://github.com/wrapperband/Feathercoin/tree/0.9.3.2/doc

Wellenreiter

@ghostlander said:

@lizhi Your patch allows to put ANY block version above 2 in new blocks. Like 102938. It defeats the purpose of block versions at all. If someone puts maliciously such a block version in a new block, we have a forked network between all existing wallets and v0.9.3.2.

https://github.com/FeatherCoin/Feathercoin/commit/911098d4b9124ff01406729efc345a4dbbff6d68

Don’t upgrade to v0.9.3.2 unless you’re ready for a trouble.

@lizhi
@ghostlamder

Do we need to accept block version 3 in 0.9.3.2?

It is an easy fix to accept block version 2 and 4 or 2,3.4 only

ghostlander

@Wellenreiter We can accept any block version if it’s labelled 2. Otherwise it needs a hard fork.

lizhi

I think 0.9.3.2 is selection of transition. It accept V2 and V4 , and create V2 only. then 0.11 create V4.
last reject Version=2 blocks when 95% of the network has upgraded.

https://github.com/FeatherCoin/Feathercoin/commit/1e2cc219d6ed147ff80a5b1fa2200d661a2e4c7c

ghostlander

@lizhi If it accepts both, someone may create a v4 block and fork the network.

wrapper

@Lizhi and @ghostlander

Couldn’t they both remain on v2 untill 95% network is v 0.11 it then auto forks to v4 ?

Then last 5 % then just need to change over.

ghostlander

@wrapper There is no way to know 95% of the network is v0.11 unless it produces somewhat different coin base.

wrapper

The only way is if Lizhi is intimating that version v4 and v2 could coexist (For example they are called v4 but act as v2 untill 95%)

It would help as we could move onto testing and updating the other feature of 0.11 before the fork.

Wellenreiter

I think, we need a set of at triggers to change to block version 4.

majority (> 80%? > 95%?) of nodes is capable to at least accept Block Version 4
- Wallets could place a comment like ‘block v4’ in the BC with each transaction if they can read V4 blocks. As not many comments -if any- are manually placed in the BC this should work
  *when receiving a block clients check for that comment
- when >xx% of blocks during a given time frame - at least 48 or 72 hours, mey be longer - meet the requirement the trigger is met.
We define a block about 6 month in the future to trigger the switch to V4 as we did with the Neoscrypt switch and announce that block number and expected date/time to the community
… ??

If you think we need other or additional triggers please comment/add…

We must make sure, that only V2 blocks are generated in the production chain until all triggers are met.
All testing must be done in the Testnet, especially the switch over must be tested upfront.
We can simulate this by using a mixture of 0.8.7, 0.9.3.1 and >= 0.9.3.2 clients in the testnet, starting with a majority of <0.9.3.2 and increasing the percentage of ‘V4 capable’ clients gradually until the trigger is met.

In parallel we could run manual checks on the seed node and the explorer to determine the client versions, but this is no guarantee, so it only can be an additional check.

I don’t like to pull any 0.11.x version to the master before we have a clear plan how we approach the switch.

lizhi

I think 0.9.3.2 accept block 2 and block 4 , but mine block 2 only . 0.11 mine block 4 only. so 0.9.3.2 is a bridge. When the main pool is installed, we will broadcast the V4.

ghostlander

@lizhi It isn’t going to work.