eth_signTypedData_v4 not supported

[anxolin]

Describe the bug
Error signing the transaction

Details
eth_signTypedData_v4 not supported.

Additional context
This comes from the feedback button, but no email was left to get more context on how to reproduce.

The message the user gets is:
the method eth_signTypedData_v4 does not exist/is not available

There's 6 methods created for historic reasons, maybe we can fallback to v3, and eth_sign as last resources?

• eth_sign
• personal_sign
• signTypedData (currently identical to signTypedData_v1)
• signTypedData_v1
• signTypedData_v3
• signTypedData_v4

It would be great to get to reproduce this.

More info in the device:

I bet is a webview within an app, probably imToken if I have to bet on one (seeing the user is from Singapore)

@nlordell WDYT?

[nlordell]

Looking at MetaMask/metamask-extension#6868 it appears that v3 and v4 work the same way, but v4 adds support for nested structs and arrays of complex types.

Since we don't make use of either of these things, I think its OK to
eth_signTypedData_v4 > eth_signTypedData_v3 > eth_sign.

Although, I think it would also be find to just eth_signTypedData_v4 > eth_sign, which would also be a bit simpler.

[alfetopito]

This is the issue we have for instance with Ledger through MM.
Also the issue I had with Ledger directly.

What we need is to add a fallback more general to switch to eth_sign whenever the EIP712 signature methods are not supported.

Right now it's hard coded only for the specific MM error code: https://github.com/gnosis/gp-swap-ui/blob/b9bc216a1b9293d2e0554d7dd6acbfd85605c102/src/custom/utils/trade.ts#L105

If the wallet is imToken as you say, we need to test and see what's the error code we get there.

[anxolin]

If the wallet is imToken as you say, we need to test and see what's the error code we get there.

Yep, it's just a guess. Im not actually sure. I say that cause i know is used is china a lot, has a webview, and uses an injected web3 instance, works similar to metamask

We should try more webviews and mobile browsers. It would be great to have a matrix with wallets we want to try out in a document, and see if we are compatible. What do you think @MareenG

We don't need to try them out in every version, but is good to test as many as we can (the ones with some market quota i mean of course)

[MareenG]

I will start such a document.
We could focus on the ones that are most used and test occasionally .

[alfetopito]

Got a report via Anna from twitter of a user having the same issue on Trust Wallet

https://gnosisinc.slack.com/archives/CPZA1AGKY/p1619745085192100
https://twitter.com/MesquitaDaniell/status/1387937258515443714?s=20

[alfetopito]

Summarizing what happened so far.
With the change on #576 the signature no longer fails due to lack of eth_sign_v4, and the order is sent to the backend.

But, the signed message is incorrect.
Here's the full slack thread https://gnosisinc.slack.com/archives/CPZA1AGKY/p1620080626246100

What was said there is basically that the recovered address from the signature is not the user's address.
Then the backend rejects the order since the random address has no funds.

With Nick's and Valentin's investigative skill, they found out that TrustWallet behaves the same as Metamask, as we have to use personal_sign instead of eth_sign.

So, the change alone in the PR (#376) won't be enough.

We need also a change to ethers.js.
I've opened a PR against their repo ethers-io/ethers.js#1542

[anxolin]

One of the users reported he used https://crypto.com/defi-wallet

EDIT: I installed the app, but i had to go through a KYC which will be done in 2-3 days

[alfetopito]

One of the users reported he used https://crypto.com/defi-wallet

EDIT: I installed the app, but i had to go through a KYC which will be done in 2-3 days

Oh boy.
I was starting to install that. I'll leave it up to you then =P

[alfetopito]

One of the users reported he used https://crypto.com/defi-wallet
EDIT: I installed the app, but i had to go through a KYC which will be done in 2-3 days

Oh boy.
I was starting to install that. I'll leave it up to you then =P

@Anxo Where where you asked for KYC? I just finished the setup and did not need any.
I think you installed the wrong app.
I know, super confusing, makes no sense, but they do have two apps with very similar names and different purpose.

You have to install the Defi wallet, which an EOA. The other app is their custodial wallet.

[alfetopito]

Regarding Status wallet -> #608

As for Crypto.com, there's no dApp browser... I mean, not in the Defi wallet. Or does the other wallet (the one you are waiting for KYC) have?

[alfetopito]

Crypto.com wallet -> #610

Doesn't work at all :(

[alfetopito]

One more to the list.
Opera (Fantom wallet).

It's basically an injected wallet provider.

Fails with the same error "The method eth_signTypedData_v4 does not exist/is not available"

It does work with the fix on #576 🥳