J J,

Actually order can be a problem.  What the configuration files do is:

Cryptography-Core
Cryptography-SHA1
Cryptography-MD5
Cryptography-DES
Cryptography-SHA256
CryptographyBase
Cryptography-ElGamal
Cryptography-RSA
Cryptography-Rijndael
Cryptography-ASN1
Cryptography-RC4
Cryptography-DSA
Cryptography-X509
Cryptography-SSL

The real issue is you should load SHA1 and MD5 before you load
CryptographyBase.  

Ron Teitelbaum


> From: J J
> Sent: Monday, October 02, 2006 11:58 PM
> 
> So, the order is now:
> 
> CryptographyBase
> Cryptography-Core
> Cryptography-DES
> Cryptography-DSA
> Cryptography-ElGamal
> Cryptography-RSA
> Cryptography-SHA1
> Cryptography-RC4
> Cryptography-MD5
> Cryptography-ASN1
> Cryptography-X509
> Cryptography-SSL
> Cryptography-DES
> 
> Right?  Or do you have to do something with versions (like download one
> version of base, then a different one later)?
> 
> >From: Robert Withers <reefedjib at yahoo.com>
> >
> >Hi,
> >
> >On Oct 2, 2006, at 2:20 PM, Hans-Martin Mosner wrote:
> >
> >>The in-image implementation existed as part of my OpenPGP efforts, and
> >>I've now ported it to the Cryptography framework.
> >>Load the MC packages CryptographyBase-hmm.18 and Cryptography-DES- hmm.3
> >>to use it.
> >
> >I tested it too, with SSL, and it works great!  Thanks!  No plugin
> needed,
> >now.
> >
> >
> >>BTW, the SSL example workspace assumes that #asURL is able to create a
> >>HttpsUrl. But that ain't so. Looking at the #asUrl code, I saw that at
> >>least in 3.8.1 this is not written with extensions in mind. Does 3.9
> >>have a more extensible method there? (I'm too tired to look myself
> right
> >>now).
> >
> >In the Cryptography-SSL Extensions, I overwrote Url
> >class>>#urlClassForScheme: to include 'https' mapped to the new  HttpsUrl
> >class.  This only works if you have SSL loaded, of course.
> >
> >I also added HierarchicalUrl>>#= and changed HTTPSocket
> >class>>#expandUrl:ip:port: to provide an ability to HttpsSocket to
> >redirect and also to detect infinite redirections, which I  experienced
> on
> >one of my test urls.  Unfortunately, I could not  update HTTPSocket
> >class>>#httpGetDocument:args:accept:request: to  take advantage.  It
> loses
> >the value of CrLf, for some reason.  I only  barely added support to the
> >HttpsSocket, so these things will change,  I'm sure.
> >
> >Cheers,
> >Robert
> >
> >
> 
>