I agree that not doing hybrid is the best option in theory, but the reality
is that some customers and jurisdictions require hybrid.

Regards,
 Rifaat


On Wed, Apr 29, 2026 at 8:51 PM David Benjamin <[email protected]>
wrote:

> Hehehe. Well, I would agree that composites are preferable to this model.
> But I also think not doing hybrids is preferable to both. :-)
>
> On Wed, Apr 29, 2026, 20:44 Andrei Popov <[email protected]>
> wrote:
>
>> Indeed. A good argument in favor of composites, even if David did not
>> intend it this way 😊
>>
>>
>>
>> Cheers,
>>
>>
>>
>> Andrei
>>
>>
>>
>> *From:* David Benjamin <[email protected]>
>> *Sent:* Wednesday, April 29, 2026 4:56 PM
>> *To:* ekr <[email protected]>
>> *Cc:* [email protected]
>> *Subject:* [EXTERNAL] [TLS] Re: anyone interested in multiple
>> CertificateVerify messages?
>>
>>
>>
>> The main challenge with doing something like that is not in TLS, but in
>> what leaks out of TLS. Systems everywhere are designed around a notion of
>> "the" peer certificate that you received over the connection. There's the
>> tls-server-end-point channel binding, but more broadly, TLS APIs tend to
>> expose a "get peer certificate" function.
>>
>>
>>
>> I think this thus falls under the "way too complicated and expensive to
>> justify doing" bucket. Once you have to move mountains to get the hybrid
>> property, it becomes extremely not worth it.
>>
>>
>>
>> On Wed, Apr 29, 2026 at 7:37 PM Eric Rescorla <[email protected]> wrote:
>>
>> Even stipulating for the moment that it's good to sign with multiple
>> certificates, I do not believe that this is the correct approach to doing
>> so.
>>
>>
>>
>> If we're going to do something here, something more like
>> https://datatracker.ietf.org/doc/draft-yusef-tls-pqt-dual-certs/ seems
>> like a better starting point.
>>
>>
>>
>> -Ekr
>>
>>
>>
>>
>>
>> On Wed, Apr 29, 2026 at 4:27 PM Stephen Farrell <
>> [email protected]> wrote:
>>
>>
>> Hiya,
>>
>> Given that it may be the case that getting certificates for
>> composite signing keys could be impractical and also involve
>> a combinatoric explosion in the number of credentials severs
>> would need to have available, I wonder if anyone has explored
>> whether it'd be useful to look at defining a way in which a
>> server (or, I guess, a client) could authenticate using more
>> than one CertificateVerify message?
>>
>> I guess that figuring that all out, and getting it implemented
>> and deployed would involve a pile of work, but ISTM it might
>> be useful, hence the question:-)
>>
>> Cheers,
>> S.
>>
>> PS: If this isn't a bonkers idea, I'd be willing to do work on
>> it, for whatever that'd be worth:-)
>>
>> _______________________________________________
>> TLS mailing list -- [email protected]
>> To unsubscribe send an email to [email protected]
>>
>> _______________________________________________
>> TLS mailing list -- [email protected]
>> To unsubscribe send an email to [email protected]
>>
>> _______________________________________________
> TLS mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>
_______________________________________________
TLS mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to