[openpgp] Re: text vs. binary in an OpenPGP "Signed Message"

[Justus Winter <[e-mail]>]

Daniel Huigens <[e-mail]> writes:

> Hi :)
>
> On Monday, March 3rd, 2025 at 10:49, Justus Winter wrote:
>> > - If not, should a verifier that encounters (d) attempt to apply CRLF
>> >   line endings to the LITb?
>> 
>> No.
>
> I'm confused by either the answer or the question :')
>
> When I answered yes, I meant, the verifier should normalize the line
> endings of the contents of the literal data packet before feeding it
> into the hash of the signature.
>
> I agree that it shouldn't affect the data returned to the application.

And Andrew writes:

> We need to be careful to distinguish between canonicalisation of the
> signature subject (i.e. the data on the wire that a receiver can read
> directly) and of the “type-specific data” that is passed into the
> signature hash function. The point of type 1 signatures is that the
> type-specific data is independent of which newline convention (if any)
> has been applied to the subject. So for the above question, perhaps we
> should clarify that the wire format of the subject SHOULD NOT be
> altered by the receiving implementation, but the type-specific data
> passed to the signature verification MUST still be normalised.

I think we are all in agreement.  Sorry if I was unclear.

Best,
Justus