Conversation
b51e019 to
00475f0
Compare
|
My commit for upgrading vergen is good to move to CodeChain. And looks ok. |
|
I think this error in https://travis-ci.com/github/CodeChain-io/codechain/jobs/368532029 is not just timeout |
MSNTCS
left a comment
There was a problem hiding this comment.
My Commits on foundry for auto self nomination is good to be included in codechain. SGTM
00475f0 to
d19564d
Compare
|
@junha1 I'll try to make the test more consistent. |
|
Ok. I'm ok with my commit |
d19564d to
01076c7
Compare
01076c7 to
ac45dc1
Compare
|
Fixed by this PR: #1982
|
sgkim126
left a comment
There was a problem hiding this comment.
codechain-stratum also uses parking_lot. You should upgrade it too. Others look good to me.
dc5aad1 to
43bb362
Compare
|
@sgkim126 I updated stratum! Thanks. |
|
@byeongjee Could you check this PR? |
Currently, we have a tool written in TypeScript to run auto-self-nominate. In this new patch, codechain is able to run auto-self-nominate in a different thread. In order to test the functionality of auto-self-nominate, a new `selfnomination` test is added in `e2e.dynval/2/selfnomination.test.ts`.
The purpose of the encoded type is to read a few fields from binary data when we don't want to full deserialize it. In the header sync, we already deserialized the whole struct in the verification process. We don't need to encoded Header when we have ctypes::Header.
We gain nothing from initializing extra data of genesis header with hash of common params, because the check will be done by the same node, and it will always succeed. We can just use extra_data written in scheme file.
CodeChain was using only the VerifierType::Canon.
CodeChain is using only CanonVerifier. We are not using the `Verifier` abstraction.
Client::import_header function receives a header that is not verified. Importer::import_headers assumes that received headers are verified. To reduce misunderstanding, I renamed the function.
The name `import_sealed_block` does not give enough context.
The h is always None.
Currently, AccountProvider in the miner is optional to support unit tests who cannot use the account provider which uses the permanent storage. This commit makes the miner uses memory back-end AccountProvider for test.
I changed the test code only. The tests were deprecated. I updated RLP encoded data from CodeChain code and it worked.
The seal is an array of any RLP data. Wrapping them with Buffer makes invalid RLP encoding.
Since Header has a cache of its hash using RefCell, the existence of the cache changes the compare result.
This will make CodeChain in tests not to increase view to high. Before this commit, some tests (for example dv.nomination.test.ts) sometimes creates a block with a very high view like 9. It was because blocks were created so fast and the timestamp of a block is too future for CodeChain.
Currently, the client access the miner through the importer. It's better to give a method to access the miner directly.
43bb362 to
62cded4
Compare
@dynaxis @sgkim126 @byeongjee @junbeomlee @MSNTCS Could you check the commits you made? I moved commits from Foundty to CodeChain. Please tell me if there are omitted commits that should be moved to CodeChain or if I moved commits that should not be moved.
I run tests on my local machine. There are some flaky tests. If I run the tests multiple times, they eventually succeed.
I've moved refactoring and bug fix PRs from Foundry to CodeChain
ctypes::Headerinstead ofencoded::Headerin HeaderDownloader Usectypes::Headerinstead ofencoded::Headerin HeaderDownloader foundry#325 @majecty