I'm pretty much on the same page as Mike here. In my fantasy world in which JOSE does not engage in protocol pontification with regard to ciphers, JWE would allow legacy, unauthenticated algorithms as well. But at the very least, we should allow JWK to represent keys for legacy algorithms, even if they can't be used with JWE.
--Richard On Sun, Nov 10, 2013 at 8:47 PM, Mike Jones <[email protected]>wrote: > I disagree with Jim and John on this. WebCrypto is a much lower-level > API than JOSE. It intentionally enables the use of raw cryptographic > primitives. For instance, when building a JOSE implementation on > WebCrypto, one would definitely use WebCrypto’s support for AES CBC to > build the composite authenticated algorithm A128CBC-HS256. Likewise, one > would likely use WebCrypto’s AES EBC support to build A128KW. > > > > The high-level API that WebCrypto is talking about **is** a JOSE > implementation. We should do everything we can to encourage that being > specified and built. > > > > Therefore, I believe we want to encourage WebCrypto to use JWKs for their > key format. Yes – even for algorithms that aren’t inherently safe. > > > > I understand Jim’s desire not to sanction the use of non-authenticated > encryption algorithms **with JWE**. That doesn’t mean that we should > prohibit all JWKs that contains keys for algorithms use them. > > > > I believe the way to do this is to expend the definition of “Algorithm > Usage Location(s)” in the JSON Web Signature and Encryption Algorithms > registry > to permit additional usage locations. Currently they must be “alg” or > “enc”. Jim has already discussed a possible extension (key management for > MAC) in which other usage locations might be used. WebCrypto could define > another usage location for their registrations. Then we could prohibit the > registration of non-authenticated encryption algorithms for “enc”, but > allow them for other cases. > > > > Would that work for people? > > > > -- Mike > > > > *From:* Mark Watson [mailto:[email protected]] > *Sent:* Sunday, November 10, 2013 7:09 PM > *To:* John Bradley > *Cc:* Jim Schaad; Mike Jones; > [email protected]; [email protected] > > *Subject:* Re: [jose] #187: Define algorithm names for symmetric keys in > for JWK > > > > John, > > > > I want to check that we have a common context here. The W3C WebCrypto API > allows Javascript content on a web page to generate / agree / import keys > which can then be used to encrypt arbitrary data, using a variety of > algorithms (including non-AEAD ones). What data and what algorithms is up > to the page developer and WebCrypto provides *plenty* of rope with which > people can hang themselves, security-wise. There is an ambition to create a > different, higher-level, API which would not provide so much rope. What > we're working on now, though, is explicitly a low-level API with all the > caveats that go with that. > > > > WebCrypto also allows keys to be imported and exported in JWK format. > > > > Are you saying that JWK is not really intended for this application ? Or > that WebCrypto should not provide support for AES-CTR and AES-CBC at all > (and if it does you want no part in supporting that with JWK) ? > > > > What about an application that wants to pass keys around in JWK format and > then use those keys to decrypt some legacy data structure encrypted with > AES-CTR ? > > > > ...Mark > > > > > > On Mon, Nov 11, 2013 at 10:45 AM, John Bradley <[email protected]> wrote: > > I agree that JOSE should not allow non-AEAD algorithms to be registered. > I understand some people will want them. > > In the words of Nancy Reagan "Just say No" I think she also said > something about your brain on non-AEAD. Who an I to argue with Nancy:) > > > > John B. > > > > > > On Nov 10, 2013, at 6:36 PM, Jim Schaad <[email protected]> wrote: > > > > Ii mean that I would like to prohibit anyone from registering a > non-AEAD algorithm. > > > > Good practice says that you should have an AEAD type algorithm for > encrypting a key so that it includes an integrity check as part of the > decryption process. Any such algorithm would qualify as an AEAD > algorithm. AES-CBC and AES-CTR do not have this property and therefore > should be prohibited from being registered and used. > > > > Jim > > > > > > *From:* Mark Watson [mailto:[email protected] <[email protected]>] > *Sent:* Sunday, November 10, 2013 5:37 PM > *To:* Jim Schaad > *Cc:* Michael Jones; [email protected]; > [email protected] > *Subject:* Re: [jose] #187: Define algorithm names for symmetric keys in > for JWK > > > > Jim, > > > > Do you mean that JOSE will not register non-AEAD algorithms in future or > that you would like to prohibit anyone from registering such algorithms ? > > > > In W3C WebCrypto we support import / export of a WebCrypto Key object in > JWK format and so I believe we will need alg / use / other attributes to > reflect all the algorithms / usages and other properties that WebCrypto Key > objects can have. > > > > ...Mark > > > > On Mon, Nov 11, 2013 at 5:30 AM, Jim Schaad <[email protected]> > wrote: > > While I agree this item is appropriately addressed as Won't Fix. I > disagree > that it would be appropriate for a later specification to define non-AEAD > algorithm for encryption purposes. If you feel it is appropriate then I > would like to make a change to the registration template to forbid it. > > Jim > > > > > -----Original Message----- > > From: [email protected] [mailto:[email protected]] On Behalf Of > > jose issue tracker > > Sent: Friday, November 08, 2013 4:46 PM > > To: [email protected]; > [email protected]; > > [email protected] > > Cc: [email protected] > > Subject: Re: [jose] #187: Define algorithm names for symmetric keys in > for > > JWK > > > > #187: Define algorithm names for symmetric keys in for JWK > > > > > > Comment (by [email protected]): > > > > A JOSE working group decision was made early on to only support > > authenticated encryption algorithms. Neither of AES CBC or AES CTR are > > authenticated encryption algorithms. > > > > There are registered algorithms for the composite AES-CBC-HMAC-SHA2 > > algorithms, which do provide authenticated encryption, which could be > used > > when applicable. > > > > That being said, it would be fine for non-JOSE specifications to define > and > > register the values A{128,192,256}CTR and A{128,192,256}CBC. For > instance, > > a W3C WebCrypto specification could do this. But I believe that JOSE > specs > > defining these values is out of scope. > > > > Therefore, I believe that this issue should be closed as "wontfix". > > > > -- > > -------------------------+---------------------------------------------- > > > -------------------------+--- > > > Reporter: | Owner: draft-ietf-jose-json-web- > > [email protected] | [email protected] > > Type: defect | Status: new > > Priority: minor | Milestone: > > Component: json-web- | Version: > > algorithms | Resolution: > > Severity: - | > > Keywords: | > > -------------------------+---------------------------------------------- > > > -------------------------+--- > > > > > Ticket URL: < > http://trac.tools.ietf.org/wg/jose/trac/ticket/187#comment:2> > > jose <http://tools.ietf.org/jose/> > > > > > _______________________________________________ > > jose mailing list > > [email protected] > > https://www.ietf.org/mailman/listinfo/jose > > > > _______________________________________________ > jose mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/jose > > > > > > _______________________________________________ > jose mailing list > [email protected] > https://www.ietf.org/mailman/listinfo/jose > >
_______________________________________________ jose mailing list [email protected] https://www.ietf.org/mailman/listinfo/jose
