- Testing
- Client Updates
- Research Updates
- Constantinople
- ProgPoW
Call starts at [5:33]
- Martin update
- Tests not yet stabilized, still activity in generating the tests
- Earlier today we found a consensus issue in the cpp-ethereum engine, this has been fixed, waiting for tests to be regenerated
- Can't speak to coverage of tests
- Dimitry update
- Getting reports of failing tests from different developers, we are fixing these issues, tests already regenerated and merged
- Hope we find all consensus issues during this test period
- But extcodehash tests and some around storage, blockchain transition tests still in dev
- Difficulty formula tests merged in past week, see
basictests/difficulty-constantinople-json
- Martin: Fuzz testing framework running for a couple of weeks, has run millions of tests, some false positives, found one consensus issue but it cannot be triggered on mainnet since it's due to behavior of non-existing precompiles
- Hudson: Have you been getting help with testing or do you still need help?
- D: No help developing tests in JSON format, Jared helped a bit
- Parity (Afri)
- Did Constantinople releases today: 2.0.7, 2.1.2
- Prepared for Ropsten, Kovan HF
- Geth (Peter)
- We did a release Monday or Tuesday which enabled all EIPs for Ropsten HF
- Trinity (Piper)
- Working on sync performance, getting sync time down to a reasonable level
- Have made a lot of progress but still a lot left to do
- EthereumJ (Dmitrii)
- Published release just now, Constantinople on Ropsten
- Not passing all tests, I hope no consensus issues
- Aleth (Pawel)
- Released v1.4, stable for some time
- Mostly an effort to have stable releases from time to time
- Not fully Constantinople-compatible
- But master branch is and we have one node running cpp on Ropsten
- Switched to parity's network monitor, so it should be on the list
- This is the only node that will participate in the Ropsten HF
- Nimbus (Jacek)
- Continuing to work on sync and getting general state tests to pass, making steady progress
- ETH2 beacon chain, work being done on test format, if you're interested, check beaconchain project on ethereum, there are open issues
- Looking into platforms, making sure we can compile on ARM and others
- Pantheon (Meredith)
- Implemented all Constantinople EIPs
- TurboGeth (not present)
- EthereumJS (not present, left comment)
- Making good progress, have substantial parts of all EIPs implemented
- Merging SSTORE implementation
- Confident we can do Constantinople ready release of VM within 1-2 weeks
- Mana (Andrew)
- Updated all EIPs, working on syncing Ropsten, began last week, at around block 300k
- Mainnet, we are at 1.7M
- Nethermind (not present)
- Exthereum (not present)
- Ewasm
- Lane
- Lots of work on benchmarking, will report results at some point
- Testnet running and mostly stable, will launch publicly at DevCon
- Working on documentation
- Pawel
- Close to release EVMC v6, need to update go bindings
- Don't want to modify more for next month
- Lane
- Hudson: Every client is totally Constantinople-compatible
- Except: Trinity, ethereumJS, exthereum
- All major clients have updated CREATE2 specs
- Danny update
- Justin working on VDF feasibility study, effort to create strong, unbiasable randomness
- Filecoin has agreed to split the cost of the next chunk of the feasibility study which involves circuit design, they are interested in it for the same reason
- Within next couple months we'll hopefully have an idea if this is going to work for the random beacon
- Randao serves our purposes for most uses, will likely move fwd with this and can add VDF as hardening in future
- Been cleaning up, expanding ETH 2.0 spec, lots of development going on
- You guys all have lots of valuable insights to add to this process
- There is an ETH2.0 specs repo, eth2.0-pm, please take a look, don't hesitate to reach out to me to talk about what's going on and where you can get involved
- Progress tracker
- Hudson
- Ropsten HF will happen around Sunday or Monday
- We'll have folks online
- If anything bad happens we'll get together
- Mainnet HF will happen after DevCon
- In about two weeks, first folks will be arriving in Prague for DevCon, should we skip the next meeting?
- (No complaints, consensus achieved, we'll do the next meeting in four weeks)
- Martin
- Everything is implemented in all clients
- If we find no consensus errors in testing or on the testnet, when should we aim to do the HF?
- Afri: mid-November might be a bit rushed
- Maybe end of November?
- Piper: second end of November
- Hudson: How much time do we need after setting the block number, before the fork?
- About a month
- When should we decide the block number?
- Piper: Can announce it outside of this call, confirm on 11/9 call
- Afri: Should we have a quick call in ~one week, just evaluate the fork on Ropsten, see if it's going well, announce a preliminary block number
- If something goes wrong, we can put it on hold until January
- Martin: How do we avoid the fiasco of the last HF?
- Peter: We discussed at the last DevCon, whether we need a bailout mechanism for the HF
- Piper: Let's take this offline and see if we can settle on something before the meeting next Friday
- Hudson: Having a smart contract that the clients can ping to see if the fork is still on - is this the idea?
- Dimitry: Clone existing mainnet, where every client repeats existing mainnet tx as if already running on Constantinople
- Martin: We could do it but no one would be using many of the changes, bomb delay not noticable, etc., would be useful to see gas changes
- Afri: Check out https://ropsten-stats.parity.io/, WS secret same as ethstats
-
Martin: Andrei, Pawel and I have been in touch w ProgPoW folks
- Have ironed out some kinks
- Some problems with the EIP
-
Pawel: Some big gaps in EIP spec
- Practically impossible to implement it based on the EIP itself
- Had to reach out for source code mostly done by ProgPoW team
- Final version found somewhere in one of the implementations, we discussed the differences
- C++/Go implementation on the same page
- EIP is in bad shape
- There are some very small improvements we can make
- Most important part is we have two reference implementations but no proper spec
- I don't like this situation as I don't want people to have to read the C or Go code to figure out how to implement it
- The team did some work on improving EIP and answering questions but not so much in public
- We are in touch with them, we can ask them questions directly and usually get some answers
- This helped me finish implementation that produced same results - just yesterday
-
Martin
- Some changes need to be made, the original idea is that it has internal epoch of 50 blocks
- When translated into ethereum it uses the same epoch as ethash which is 30k blocks, which leads to some problems
- Needs an update of protocols between the miners, the getwork function
- Other than that, the general idea is to launch a testnet for it ASAP
-
Hudson: So we want the EIP updated to include more detail
- Have we communicated this to them?
- Pawel: Yes. As of today they still plan to update the EIP with what we discussed privately, to fill the gaps.
-
Hudson
- Skipping next call
- Constantinople special call in a week
- Won't shift overall meeting schedule
- Meredith Baxter (ConsenSys/PegaSys/Pantheon)
- Paweł Bylica (EF/aleth)
- Jason Carver (EF/python/trinity)
- Dmitry (Harmony)
- Daniel Ellison (ConsenSys/LLL)
- Matthew English (ConsenSys/PegaSys)
- Andrew Gross (Mana)
- Fredrik Harrysson (Parity)
- Hudson Jameson (EF)
- Mikhail Kalinin (EF/EthereumJ)
- Eric Kellstrand (ConsenSys/PegaSys)
- Dimitry Khokhlov (EF/testing)
- Piper Merriam (EF/python/trinity)
- Lane Rettig (Ewasm)
- Danny Ryan (EF/research)
- Afri Schoeden (Parity)
- Jacek Sieka (Status/Nimbus)
- Martin Holst Swende (EF/geth/security)
- Péter Szilágyi (EF/geth)