[openpgp] Re: OpenPGP grammar concerns (fwd: [Peter Gutmann] Re: [oss-security] Retrospective by 'gpg.fail' authors)

[Andrew Gallagher <[e-mail]>]

On 06/10/2026 10:08, Peter Pentchev wrote:
> 
>> Do you have a reference? I’m not familiar enough with his works to find the specific quote.
> 
> He linked to that in his previous message: https://news.ycombinator.com/item?id=46404339
Thanks!

So he's comparing with SSLv2 not having a fully-covered handshake. Yes, 
there are some small parts of the data stream that aren't covered by 
signatures, such as the literal data metadata (which RFC9580 recommends 
should not be used).

But maybe he's thinking more about how efail took advantage of lax 
grammar enforcement at the receiver, such as the case where there were 
two literal data packets in a single message, one signed and one not. It 
is tempting to blame this on poor implementation, as it has never been 
valid grammar, but it's true the design makes this kind of 
implementation error easier than it could be.

A