Skip to content

Correct wrong bytevector verify branch - #54

Open
nicolas-graves wants to merge 1 commit into
artyom-poptsov:masterfrom
nicolas-graves:bugfix
Open

Correct wrong bytevector verify branch#54
nicolas-graves wants to merge 1 commit into
artyom-poptsov:masterfrom
nicolas-graves:bugfix

Conversation

@nicolas-graves

Copy link
Copy Markdown
Collaborator
  • modules/ssh/key.scm (verify): An unchecked copy-paste made its way in the bytevector branch of the verify function, correct it.

  • tests/key.scm ("sign: RSA, bytevector data") ("verify: RSA, bytevector data, valid signature")
    ("verify: RSA, bytevector data, invalid signature") ("verify: RSA, string data signed, verified as bytevector"): Add regression tests (they fail without the fix).

Disclaimer: tests are generated with the help of gen AI, but I read them and tested them properly.

@artyom-poptsov

Copy link
Copy Markdown
Owner

Hello!

Thanks, I'll look into it later this week.

I noticed that you used Conventional Commits notation in the commit title. I'm using Conventional Commits style at my work (and in some of personal projects), and I'm open to the idea of using this style in Guile-SSH as well. However I think that we need to keep the style consistent, so it is better to agree on the style first and then migrate to it.

Maybe we can discuss this in a separate issue? I can create a Conventional Commits migration proposal.

Also I would not add a dot at the end of the commit title.

Disclaimer: tests are generated with the help of gen AI, but I read them and tested them properly.

I'm not against AI (or, to say it better, LLMs) per se, but we need to make sure that we have throughout understanding of what the code does and how it does that; as a rule of thumb we have to be able to write the same code ourselves without help of an LLM and argue about the code fluently.

Personally I see LLMs mostly as "brain augmentation" rather than "brain substitution".

@nicolas-graves

nicolas-graves commented Aug 11, 2026

Copy link
Copy Markdown
Collaborator Author

I noticed that you used Conventional Commits [...] Maybe we can discuss this in a separate issue? I can create a Conventional Commits migration proposal.

I created a dedicated issue #57

I'm not against AI (or, to say it better, LLMs) per se, but [...] Personally I see LLMs mostly as "brain augmentation" rather than "brain substitution".

FYI that is also quite close to my own opinion. I would also add that the projects should be able to be maintainable without the most advanced models (so not only expertise, but also churn limited to what's feasible without paying.

* modules/ssh/key.scm (verify): An unchecked copy-paste made its way in the
bytevector branch of the verify function, correct it.

* tests/key.scm ("sign: RSA, bytevector data")
("verify: RSA, bytevector data, valid signature")
("verify: RSA, bytevector data, invalid signature")
("verify: RSA, string data signed, verified as bytevector"):
Add regression tests (they fail without the fix).
@nicolas-graves nicolas-graves changed the title fix(verify): Correct wrong bytevector verify branch. Correct wrong bytevector verify branch Aug 11, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants