[Jsdsi-devel] Re: jsdsi.util.KeyFactory

[Sean R. <sra...@ae...>]

Sameer Ajmani wrote:

>Sean,
>
>In answering Will's question, I noticed that jsdsi.util.KeyPairFactory
>is does not implement a standard Java interface.  Instead, it offers
>myriad create() routines and keeps and internal cache of
>KeyPairGenerators.  Why is this?  Perhaps we discussed this when you
>first implemented this class, but now I think there's a cleaner way to
>do it:
>
>Instead let's have a class jsdsi.util.RSAKeyPairGeneratorSpi that
>implements java.security.KeyPairGeneratorSpi.  Instead of having so
>many different create() routines, the parameters for creation are
>passed via java.security.KeyPairGeneratorSpi.initialize(...).  The
>parameter to initialize() is an instance of a new class,
>jsdsi.util.RSAKeyGenParameterSpec, which is a subclass of 
>java.security.spec.RSAKeyGenParameterSpec; our subclass contains extra
>fields for URIs, etc.
>
>With this class, users can generate KeyPairs in the standard Java way:
>KeyPairGenerator kpg = KeyPairGenerator.getInstance("RSA", new
>jsdsi.Provider()) ;
>kpg.initialize(new RSAKeyGenParameterSpec(keysize, exponent, uris));
>KeyPair kp = kpg.genKeyPair();
>assert(kp.getPublic() instanceof jsdsi.RSAPublicKey);
>assert(kp.getPrivate() instanceof java.security.interfaces.RSAPrivateKey);
>
>I believe jsdsi.util.RSAKeyPairGeneratorSpi will be somewhat easier to
>maintain than  jsdsi.util.KeyPairFactory and will be more useful to
>users.  I apologize for not thinking of this long ago :)
>
>If you have a good reason for why jsdsi.util.KeyPairFactory is better,
>please let me know!
>  
>
I can't recollect myself. The only thing I remember was the issue with 
the Sun implementation giving the same key pairs on consecutive calls 
when using the 'defaults'. I'll look into it this, or next week.

Sean

-- 
Dr. Sean Radford, MBBS, MSc
sra...@ae...
http://www.aegeus-technology.com/