[cabfpub] Fwd: Re: Proposal to add DSA 2048

Adam Langley agl at google.com
Mon Mar 11 07:11:01 MST 2013


On Mon, Mar 11, 2013 at 6:32 AM, Erwann Abalea
<erwann.abalea at keynectis.com> wrote:
> It may be worth mentioning one other thing beyond Erwann's summary about DSA keys that is unlike RSA: that only part of the public key may be specified in the certificate, and it is expected to inherit the missing parameters from the parent certificate (or fail).  This is an odd sharp edge that came up in Public Key Pinning - and I'm sure it will cause some applications somewhere to crash ;)

I haven't seen Symantec's DSA certificates yet, but I very much hope
that they don't do this. As I recall, this behaviour may be tested by
PKITS and so may actually be supported in some certificate
verification libraries. None the less, the world will be a simpler and
better place if this corner of PKIX never sees the light of day. (And
that hope may end up being expressed in code.)


Cheers

AGL


More information about the Public mailing list