Add checksum to getdescriptorinfo

[sipa]

No description provided.

[gwillen]

So the only reason I didn't write it this way, is that people have said they don't want private keys to be echoed in the output.

Personally I think this is fine -- the only way to get a private key in the output, is if you include it in the input. And in fact, there has to be SOME way to get private keys echoed back, if you want to be able to get checksums for descriptors with private keys included.

So I'd advocate for this PR the way it is, but if people aren't comfortable with it, I'd suggest adding a flag, and not echoing descriptors with private keys unless the flag is set.

Thanks!

(Tagging #15740 for the record.)

[meshcollider]

utACK 15c482e

[Sjors]

Concept ACK.

Light preference for returning only the checksum rather than echoing a private key. But I do agree with @gwillen's point:

the only way to get a private key in the output, is if you include it in the input.

I find the extra with_checksum confusing. Maybe we can add an argument keep_private_keys which defaults to false. I would also prefer sticking to canonical form (but with private keys). An additional argument canonical can be true by default. I know it makes the change a bit more complicated, but it seems more robust in the long run.

[DrahtBot]

The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.

Conflicts

Reviewers, this pull request conflicts with the following ones:

• #16542 (Return more specific errors about invalid descriptors by achow101)

If you consider this pull request important, please also help to review the conflicting pull requests. Ideally, start with the one that should be merged first.

[jb55]

I'm confused by this, getdescriptorinfo already adds a checksum if it's missing?

[sipa]

@jb55 It also normalizes the descriptor (which among other things turns WIP into pubkey, and xprv into xpub)

[jb55]

ah ok that was the bit I was missing

[promag]

ACK 15c482e, only reviewed code change. Looks like returning only the checksum, like @Sjors suggested, is a simple solution to not echo back private key. A small release note could be added.

Refactor commit best viewed with git show --color-words 7976dd65f6503082370a40ac189b925322c29918.

[achow101]

Code Review ACK 15c482e

I would prefer if there were also a test where the descriptor were slightly modified (e.g. replacing ' with h in origin info) and that the result of getdescriptorinfo gives the correct canonical form and the modified form with the checksum.

[Sjors]

Tested ACK 15c482e

[sipa]

I like the suggestion of instead just reporting the checksum for the unmodified descriptor, and not echoing anything back. Since this has been reviewed already, I'm happy to go either way. What do reviewers think?

[achow101]

I like the suggestion of instead just reporting the checksum for the unmodified descriptor, and not echoing anything back. Since this has been reviewed already, I'm happy to go either way. What do reviewers think?

I prefer this because it is unambiguous as to what the checksum belongs to and what the descriptor with with checksum will look like.

[sipa]

Updated to instead add a "checksum" to getdescriptorinfo that returns the checksum for the (unmodified) descriptor input.

[Sjors]

ACK 26d3fad

[Sjors]

nit: did you really mean chksum?

[achow101]

Code Review ACK 26d3fad

[fanquake]

Two wallet reviewer ACKs here; @meshcollider want to take a final look? I've compiled to check that there aren't any silent merge conflicts etc.

[meshcollider]

re-Code Review ACK 26d3fad