access list test example [example]

[winsvega]

Retesteth: ethereum/retesteth#114
Geth: ethereum/go-ethereum#21502

[holiman]

Reposting this here: I think the "data" : ":label somelabel :raw 0x1122", is really not a good idea. I think it's fine to have

 "data" : "0x..",
 "_dataCommment": "Foo bar bazonk",
 "accessList" : [ ...] 
"_anotherComment": "Baz bar",

But inserting logic into the data-field is really making things too complex, and doesn't play well with json-parsers: e.g. in golang, our parsing is typed. As in, we define what the type of a specific input is, and having some mixed, loosely defined meta-format with :-delimited fields just adds another level of quirkyness

[MariusVanDerWijden]

I don't really know how to unmarshal this properly as the key depends on the length of data. I would suggest to drop the index and just make accessLists an array like data with the same length and leave elements empty if you don't want to have an access list for it. cc @winsvega @holiman

e.g.

type stTransaction struct {
	GasPrice    *big.Int       `json:"gasPrice"`
	Nonce       uint64         `json:"nonce"`
	To          string         `json:"to"`
	Data        string         `json:"data"`
	AccessLists []types.AccessList `json:"accessLists,omitempty"`
	GasLimit    []uint64       `json:"gasLimit"`
	Value       []string       `json:"value"`
	PrivateKey  []byte         `json:"secretKey"`
}

type AccessList []AccessTuple  `json:"omitempty"`

type AccessTuple struct {
	Address     *common.Address `json:"address"        gencodec:"required"`
	StorageKeys []*common.Hash  `json:"storageKeys"    gencodec:"required"`
}

[MariusVanDerWijden]

So basically my idea would be something like this:

"transaction" : {
            "data" : [
                "0x1122",
                "0x3344",
                "0x3234"
            ],
            "accessLists" : [
                    {
                        "address" : "0x0000000000000000000000000000000000001337",
                        "storageKeys" : [
                            "0x0000000000000000000000000000000000000000000000000000000000000001",
                            "0x0000000000000000000000000000000000000000000000000000000000000002"
                        ]
                    },
                    {
                    
                    },
                    {
                        "address" : "0x0000000000000000000000000000000000001337",
                        "storageKeys" : [
                            "0x0000000000000000000000000000000000000000000000000000000000000001",
                            "0x0000000000000000000000000000000000000000000000000000000000000002",
                            "0x0000000000000000000000000000000000000000000000000000000000000002"
                        ]
                    },
            ],
            "gasLimit" : [
                "0x061a80"
            ],
            "gasPrice" : "0x01",
            "nonce" : "0x00",
            "secretKey" : "0x45a915e4d060149eb4365960e6a7a45f334393093061116b197e3240065ff2d8",
            "to" : "0x095e7baea6a6c7c4c2dfeb977efac326af552d87",
            "value" : [
                "0x0186a0"
            ]
        }

[holiman]

I agree about that. To clarify: {} can be used for empty list, and null to signify that this is a legacy-tx.

[holiman]

I don't really know how to unmarshal this properly as the key depends on the length of data.

Unmarshalling shouldn't be a problem, though, it's just a map[int]AccessList. But it's still a bit convoluted

[winsvega]

you think just null for the transaction with data where accessList not present is better ?

ok. it is just the matter of test format. which one is easier for you guys.

[winsvega]

I removed the map and now it is just an array (same length as data array)
{} for absense of accessList
[] for empty accessList

ori alsomst finished his tests and will generate soon on this branch
#774

[winsvega]

Is there a problem with parsing {}?

Martin wants it to be explicit null

[holiman]

Yes, the problem is I'm parsing the input into a list type. When it reads an object-type, the parsing fails. Null should be fine, but mixing types is not easy to handle.

[winsvega]

added "txbytes" field in gstate post section. so that no need to build a transaction and do the signing operation from transaction section (complicated logic)

here also a blockchain test example with access list