Thomas Roessler writes:
Thus, Ulf's question remains valid: If you follow the old behaviour
in one of the more important fields of application anyway, why on
earth didn't you make sure RFC 2440 and your applications followed
this kind of behaviour in general? After all, he had indeed warned
of the inconsistency introduced by RFC 2440.
It is true, Ulf posted on this twice, on 16 Sep 98:
| 0x01: Signature of a canonical text document.
| Typically, this means the signer owns it, created it, or
| certifies that it has not been modified. The signature is
| calculated over the text data with its line endings converted to
| <CR><LF> and trailing blanks removed.
Trailing blanks are removed _only_ for cleartext signatures, in the
transformation described in section 7.1. Otherwise (for signatures
contained in a PGP message, and for detached signatures) they are part
of the signature.
BTW, there's no proper definition of detached signatures in the draft.
They probably should be mentioned in section 10 because they are used
for PGP/MIME.
and again on 13 Oct 98:
Looks good, but the definition for the canonical text signature is
still wrong. (Also I note you didn't include the note about PGP 2.6.x
being unable to hande signature and pubkey packets of length type 0.)
| 0x01: Signature of a canonical text document.
| Typically, this means the signer owns it, created it, or
| certifies that it has not been modified. The signature is
| calculated over the text data with its line endings converted to
| <CR><LF> and trailing blanks removed.
Trailing blanks are removed _only_ for cleartext signatures, in the
transformation described in section 7.1. Otherwise (for signatures
contained in a PGP message, and for detached signatures) they are part
of the signature.
The RFC issued on 11 Nov 98 without this change.
I can't account for our overlooking these two comments. It was an error
on my part and that of the other authors. No process is perfect.
I note that the current RFC2440bis draft still has the incorrect language.
Should we change the new draft to document the way the commercial versions
of PGP, and versions 2.X, have always worked?
Hal