Reuse same type parsing for defaultAbiCoder.encode as in utils.Interface

[snario]

If I have

const ENCODING = `
  tuple(
    address foo,
    bytes4 bar
  )`;

Then the following will work:

new ethers.utils.Interface([ENCODING]).encode({ foo, bar });

However the following will error:

ethers.utils.defaultAbiCoder.encode([ENCODING], [{ foo, bar }]);

Like so:

    unexpected character "a" at position 22 in "
      tuple(
        address foo,
        bytes4 bar
      )"

I think it is because different parsing logic is used? @ricmoo

[ricmoo]

I think this is because I don’t handle newline as whitespace... I can look into it. I can probably just replace \n with space...

[ricmoo]

Oh. Looking at the above code again, you are using Interface, which expects a function of sent signature, not a type.

In the first example it works in the first case because you are constructing a function named tuple, which takes in an address and a bytes4. It has nothing to do with an actual tuple.

I have confirmed the issue you are experiencing is the newlines. For now, if you use ENCODING.replace(/\n/g, '') it will work fine. I will look into treating those as normal whitespace for the aBI coder.

[snario]

Ah, makes sense, thanks @ricmoo. It remains to be the case that it works for this, too:

new ethers.utils.Interface([`foobar(${ENCODING})`]).encode({ foo, bar });

[ricmoo]

You should be able to safely use newline (or any other whitespace) now in the human-readable ABI, in version4.0.16. Let me know if there are still problems, and feel free to re-open this issue.

Thanks! :)