Add arrowGlacier HF

[jochem-brouwer]

This PR adds the arrow glacier HF, which delays the difficulty bomb.

Since we have no test cases, do not merge yet.

[codecov]

Codecov Report

Merging #1527 (7807f02) into master (c230e51) will not change coverage.
The diff coverage is n/a.

Flag	Coverage Δ
block	84.96% <ø> (ø)
blockchain	85.08% <ø> (ø)
common	100.00% <ø> (ø)
devp2p	82.89% <ø> (ø)
ethash	∅ <ø> (∅)
tx	87.44% <ø> (ø)
vm	79.08% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

[holgerd77]

For reference: ethereum/tests#965 (ethereum/tests tracking issue for test implementation)

[holgerd77]

For reference: ethereum/tests#972 (new difficulty tests on ethereum/tests)

[holgerd77]

I guess at some point we should also get over the finish line with this. Any updates here? What's the current status? 🤔

[jochem-brouwer]

If we want we can merge this and then at a later point run it against ethereum/tests? I can try running the test locally to see if they pass, if thats the case then we can also merge.

[holgerd77]

Ok, maybe let's then wait for the official ethereum/tests release, guess that can't be too far away, and some in between step we should also update to v10.1 since this release comes with some significant structural changes (legacy tests moved to own repo).

@jochem-brouwer Can you do rebase in between and bring this up-to-date so that we can keep up here a bit with keeping this close to master?

[holgerd77]

Will mainnet be the only fork here or will there be a test fork on ropsten? Or would this not working out due to differing practical TD values anyhow? 🤔

[jochem-brouwer]

Ok this is weird, the tests do not pass, I tried locally. To test;

cd packages/ethereum-tests
git checkout develop
git pull

This pulls the latest data from the repo, the difficulty tests are not added yet.

Now to /block/test/difficulty.spec.ts add this to hardforkTestData (line 37)

arrowGlacier: require('../../ethereum-tests/DifficultyTests/dfArrowGlacier/difficultyArrowGlacier.json').difficultyArrowGlacier.ArrowGlacier

Only running these tests;

1..10776
# tests 10776
# pass  9896
# fail  880

[holgerd77]

Yes, can confirm.

Also ran this locally, when digging a bit deeper one sees that every second test with "parentUncles" : "0x00" is failing.

Not sure yet what is happening, the format of this uncles field changed substantially (there was a full hash before), will ask on ethereum/tests.

[holgerd77]

The semantics of the uncles field has changed, see here.

If one is replacing 0x00 (which now stands for "no uncles") with an just an empty field (which translates for our instantiation to "instantiate with no uncles", this works again.

Will push.

[jochem-brouwer]

@holgerd77 I see, but I don't understand how this passes. I don't see anything in your commits which handles this parentUncles field, so I do not understand why this passes. Is it an edited tests file?

(Small sidenote: we should at some point make these block tests not use the copied JSON files from ethereum/tests, but directly read them from ethereum/tests)

[holgerd77]

Yes, edited test file.

[jochem-brouwer]

Ah right. We should fix that at some point, I don't think it is so hard, seems like the difficulty formula is adjusted if either the uncle hash is the empty RLP list hash OR it is not, so we could just create a bogus hash which should (?) not crash the difficulty calculation. Will approve here.

[jochem-brouwer]

Oh right, I can't approve my own PR 😂

[holgerd77]

LGTM and x-approved by @jochem-brouwer, will merge.