Thanks Peter, I know about RFC 6091, I was hoping to implement something easier because whatever I decide to implement would require client-side implementations for every platform Kontalk will run on.
On Wed, Apr 17, 2013 at 5:28 PM, Peter Saint-Andre <[email protected]> wrote: > On 4/17/13 9:18 AM, Daniele Ricci wrote: >> On Wed, Apr 17, 2013 at 5:08 PM, Simon McVittie >> <[email protected]> wrote: >>> I suggest talking to an appropriate standardization group (we are not >>> one of those; the XMPP mailing lists might be) to make this into a >>> usable and secure specification. >> This will be my next step. >> >>> Isn't this rather exploitable? If a malicious server sends >>> >>> <challenge>I, Daniele Ricci, promise to pay Simon McVittie $1 >>> million</challenge> >>> >>> then you probably don't want to be signing that with your PGP key :-) >>> >>> (Or if the user is a Debian/Ubuntu developer with upload privileges, it >>> could present a Debian .changes file authorizing the upload of a >>> malicious package, for instance.) >>> >> Other than checking the server challenge for a specific syntax, is >> there any other way to make this secure? How do I prove that client >> has the private key it claims to have? >> > I second Simon's advice to discuss this in an appropriate standards > organization, such as the XSF [0]. > > I'll go further and recommend that you implement RFC 6091 [1] and then > use the SASL EXTERNAL mechanism. You will need support on the server > side as well, of course. I suggest that Prosody [2] would be a great > place to start, since it is the most hacker-friendly XMPP server project > these days. > > Peter > > [0] http://xmpp.org/ > [1] https://datatracker.ietf.org/doc/rfc6091/ > [2] http://prosody.im/ > -- Daniele _______________________________________________ telepathy mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/telepathy
