Benchmark command fail due to json deserialize error

[NingLin-P]

Is there an existing issue?

• I have searched the existing issues

Experiencing problems? Have you tried our Stack Exchange first?

• This is not a support question.

Description of bug

The benchmark command now requires the genesis-builder-preset arg to get JSON-encoded RuntimeGenesisConfig from the runtime but this value is not used directly, instead it is used to patch the default value of RuntimeGenesisConfig and the result is used:

polkadot-sdk/substrate/client/chain-spec/src/genesis_config_builder.rs

Lines 153 to 155 in 01936b3

	let mut config = self.get_default_config()?;
	crate::json_patch::merge(&mut config, patch);
	self.get_storage_for_config(config)

within json_patch::merge if the value is null the whole kv pair will be removed from the JSON:

polkadot-sdk/substrate/client/chain-spec/src/json_patch.rs

Lines 36 to 38 in 01936b3

	for (k, v) in b {
	if v.is_null() {
	a.remove(&k);

this causes issue for enum that contains Option value:

enum Test {
    V(Option<u32>),
}

Test::V(None) will be serialized to {v: null}, after json_patch::merge it became { } and is impossible to deserialize back to the same value

Steps to reproduce

Add this unit test to substrate/client/chain-spec/src/json_patch.rs and it will fail due to deserialize

#[test]
	fn test_null_value() {
		use serde::{Deserialize, Serialize};

		#[derive(Debug, Serialize, Deserialize)]
		enum Test {
			V(Option<u32>),
		}

		let v1 = serde_json::to_value(Test::V(None)).unwrap();
		let mut v2 = serde_json::to_value(Test::V(None)).unwrap();

		merge(&mut v2, v1);

		let j = serde_json::to_string(&v2).expect("serialize should succeed");

		serde_json::from_slice::<Test>(&j.into_bytes()).expect("deserialize should succeed");
	}

[bkchr]

CC @michalkucharczyk

[michalkucharczyk]

Thanks for reporting and in-depth analysis of the problem. It all sounds good, removal of keys in the base JSON object which is default is indeed not necessary. The fix proposal is here: #6382

[michalkucharczyk]

would you confirm fix does work for you?

[NingLin-P]

The fix does works, thanks!