On Wed, Dec 28, 2005 at 01:34:56PM -0800, Jon Callas wrote:
Which is not to say I care passionately about reversability. I would,
however, like to remove the ambiguity.
Can we say implementations SHOULD always add a newline, but they MAY
omit it if the plaintext ends with a newline (for backwards
compatibility)? That way nothing is broken, but future behaviour is
defined.
I'm happy to make that change. In fact, I just did.
Forgive me, please, but I understood the rule before. With this
change, I don't think I understand what is expected any longer. Is
this added newline hashed into the signature?
Wouldn't "always" adding a newline cause text (with no line ending)
to become:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
text
-----BEGIN PGP SIGNATURE-----
xxxxxx
-----END PGP SIGNATURE-----
and text\n (with a line ending) to become:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
text
-----BEGIN PGP SIGNATURE-----
xxxxxx
-----END PGP SIGNATURE-----
If someone could post an example of what is now expected for "text"
and "text\n", I'd appreciate it.
GPG never adds a final newline to the hashed data, even if the
original document doesn't have one. If there is a final newline in
the document, it is removed and not hashed into the signature.
Does this change to the draft make GPG noncompliant?
David