Thanks for this!  It found some issues where my parser wasn't right.  Claude 
didn't like one of your vectors, and found some cases where your validator 
wasn't strict enough either.

  - tags_whitespace is skipped as a defective upstream vector — its .signed has 
an empty s= signature value, so pass is unachievable by anyone.
  - A few status-code notes remain (we return fail where the vectors say 
permerror: algorithm_misnamed, domain_below_mailfrom, nonce_too_long, 
mailfrom_needs_brackets) — same accept/reject decision, different DKIM status 
label.

It's really useful to have different implementations to test against, without 
the same blind spots and assumptions that I baked into mine!

Bron.

On Thu, Jun 25, 2026, at 09:10, Steve Atkins wrote:
> 'Afternoon all
> 
> I’ve been working on some test vectors that exercise bits of the spec that 
> aren’t just “this is the output of our signer, it should pass”. Things like 
> duplicate tags, tags in an unexpected order, invalid nonces, excessively 
> large or small keys, case insensitive matching and so on.
> 
> https://forge.turscar.ie/turscar/dkim2tests/src/branch/main/tests has the 
> tests themselves and 
> https://forge.turscar.ie/turscar/dkim2tests/src/branch/main/README.md 
> explains the structure of the files.
> 
> There’s no coverage of recipes or any other multihop tests - I’ll look at 
> those once I’ve updated to spec-03.
> 
> Additionally, I’ve put some online tools up at https://dkim2.eu/ that 
> complement the work Bron has been doing, letting folks test and explain their 
> signed messages.
> 
> Cheers,
>   Steve
> 
> _______________________________________________
> Ietf-dkim mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
> 

--
  Bron Gondwana, CEO, Fastmail Pty Ltd / Fastmail US LLC
  [email protected]

_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to