Should `extract_adaptor` API return `None` instead of `False`?

[siv2r]

The extract_adaptor API returns False when it’s unable to generate a valid adaptor point. In one of our calls, we decided on using False because None is used to represent an infinity point.

Wouldn’t it be more pragmatic to use None instead?

Here are my reasons:

• extract_adaptor is supposed to return a point, so returning a boolean value seems unusual
• The adaptor can never be an infinite point. At least, our APIs don’t support it, so reserving the None keyword doesn’t make sense.
  • Let’s say the user encodes the infinity adaptor point as 0x0000...0000 (33-byte array) and tries to create a pre-signature for it. This won't work because cpoint(adaptor) would raise value error. So, the adaptor can never be an infinity point in our APIs.
  • This also implies that the secadatpor can never be 0. Maybe we can add an assert for this in extract_secadaptor
  • If we really want to support signing with an infinity adaptor point, we can use cpoint_ext instead of cpoint
• BIP340 also follows this None approach. The lift_x function returns None when it cannot generate a valid point. Even though None represents the infinity point for the point_add, point_mul, and point_negate functions.

[jonasnick]

Agree that the return type should be Optional[PlainPk]. Since None is not not a plain public key, I don't see how a user could confuse this.

I think the function returns False now because it used to return Point at some point in the past and not PlainPk.