Skip serializing blob_schedule before Fulu

[michaelsproul]

Issue Addressed

Alternative to:

• fix: allow BLOB_SCHEDULE: [] field #7758

Proposed Changes

Serve the blob_schedule field on /eth/v1/config/spec only when Fulu is enabled. If the blob schedule is empty, we will still serve it as [], so long as Fulu is enabled.

Additional Info

This touches the same code as:

• Simplify ConfigAndPreset #7777

But this PR is higher prio for Fusaka testing.

[mergify]

Some required checks have failed. Could you please take a look @michaelsproul? 🙏

[pawanjay176]

LGTM, fixed the test and merged unstable

[pawanjay176]

Actually this is quite annoying to deal with in the tests because a roundtrip ser/de isn't consistent. i.e. this test would fail

#[test]
    fn chain_spec_roundtrip() {
        let yamlconfig =
            ConfigAndPreset::from_chain_spec::<MainnetEthSpec>(&ChainSpec::mainnet(), None);

        let serialized = serde_yaml::to_string(&yamlconfig).unwrap();
        let deserialized: ConfigAndPreset = serde_yaml::from_str(&serialized).unwrap();
        assert_eq!(yamlconfig.config().blob_schedule, deserialized.config().blob_schedule);
    }

Not sure of a clean way to handle this other than setting fulu_fork_epoch all places where we do this kind of roundtrip.

[michaelsproul]

Why doesn't it round trip? Is it because of the serialisation field? Maybe we should ignore that in the PartialEq impl of BlobSchedule

[jimmygchen]

Why doesn't it round trip? Is it because of the serialisation field? Maybe we should ignore that in the PartialEq impl of BlobSchedule

Yeah that's right
When we deserialise (result below), the skip_serializing value is set to false (default),
but when we create it from ChainSpec (expected), the skip_serializing value is set to true because Fulu is unset.

lighthouse/validator_client/http_api/src/test_utils.rs

Lines 257 to 265 in 257d270

	let result = self
	.client
	.get_lighthouse_spec::<ConfigAndPresetFulu>()
	.await
	.map(|res| ConfigAndPreset::Fulu(res.data))
	.unwrap();
	let expected = ConfigAndPreset::from_chain_spec::<E>(&E::default_spec(), None);
	assert_eq!(result, expected);

Yeah i think ignoring that field in PartialEq sounds good. Will push a commit now.

[pawanjay176]

Ohh ignoring it in the partialeq impl is very clean, nice!