Hi Werner,

When I was going through the sample messages to figure out
#encryptedKeySHA1 stuff I stumbled upon a potential problem with our
WSS-1.1 support.

Scenario 4.4 (no derived keys) of this [1] (Not sure whether these are
standard scenarios) seems to expect the request [2] message to be
encrypted _and_ signed (HMAC-SHA1) by an ephemeral key, and a
supporting signature over the primary signature with the requester's
private key. Also the response [3] is using the same #EncryptedKeySHA1
reference value on both signature and encryption. If we are to support
this sort of scenarios we will have to find a way to share the keys
across the builders. If we have such a mechanism we can avoid going
for an "all-in-one" approach and stick to simpler separate builders.
:-)

Thoughts?

Thanks,
Ruchith

[1] http://mssoapinterop.org/ilab/WSSecurity/WCFInteropPlugFest_Security.doc
[2] http://rafb.net/paste/results/Xq7tsn14.html
[3] http://rafb.net/paste/results/xbNh5i37.html

On 2/6/06, Dittmann, Werner <[EMAIL PROTECTED]> wrote:
> Ruchith,
>
> I didn't saw that problem. In that case it seems to be ok if
> it is "all-in-one" :-).
>
> Regards,
> Werner
>
>
> > -----Ursprüngliche Nachricht-----
> > Von: Ruchith Fernando [mailto:[EMAIL PROTECTED]
> > Gesendet: Montag, 6. Februar 2006 13:35
> > An: Dittmann, Werner
> > Cc: [email protected]
> > Betreff: Re: DerivedKey Encryption and Signature
> >
> > Werner,
> >
> > sure no problem ... please check in your changes... Will pick
> > up from there.
> >
> > Let me think about having a separate class a little bit more... the
> > only issue I had was handling the case where we have to use the same
> > ephemeral key to derive key to do both encrypt and signature. How do
> > we share the ephemeral key to be used with the signature builder and
> > the encryption builder?
> >
> > Thanks,
> > Ruchith
> >
> > On 2/6/06, Dittmann, Werner <[EMAIL PROTECTED]> wrote:
> > > Ruchith,
> > >
> > > I've refactore your DK implementation according to the model
> > > of the other encrypt/sign classes. It works perfectly with
> > > your test case. Should I check it in? Because its quite some
> > > code reshuffling this could disturb some ongoing coding on your
> > > side.
> > >
> > > Having checked your implementation I would propose to have the
> > > Signature with derived key in a separate class. IMHO otherwise
> > > the combined class may become too complex. HAving it separate
> > > would also enable us to look for common method/functions that we
> > > can combine in another base class or something similar.
> > >
> > > Ideas?
> > >
> > > Regards,
> > > Werner
> > >
> > >
> > > > -----Ursprüngliche Nachricht-----
> > > > Von: Ruchith Fernando [mailto:[EMAIL PROTECTED]
> > > > Gesendet: Montag, 6. Februar 2006 11:32
> > > > An: Dittmann, Werner
> > > > Cc: [email protected]
> > > > Betreff: Re: DerivedKey Encryption and Signature
> > > >
> > > > Werner,
> > > >
> > > >
> > > > Yes I noticed your changes and the new prepare() and
> > build() methods.
> > > > Nice work !!! I have tried to follow the same pattern in the
> > > > WSSecDKSignEncrypt by doing the EncryptedKey creation in
> > the prepare()
> > > > method and in the build() method I created the DKT and
> > the reference
> > > > list and added them to the doc . This way certainly geives us more
> > > > control.
> > > >
> > > > +1 on refactoring to support the token placement rules
> > (lax, strict,
> > > > etc ?) as for sec policy stuff.
> > > >
> > > > Thanks,
> > > <SNIP ... SNAP >
> > >
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to