CryptoBinary and base64Binary clarification

I'm a little confused about CryptoBinary and base64Binary.

As I understand it, CryptoBinary was originally defined before base64BInary
was added to schema. CryptoBinary was defined to include stripping of
leading 0 octets - providing a form of compression. When base64Binary was
added to schema, it was defined without this stripping. I believe that is
now the only difference.

What is confusing is knowing when to use CryptoBinary and when to use
base64Binary. Apparently you must use base64binary whenever a value must be
exact - e.g. a signature value, a digest value or a ciphervalue. Is that
correct?

Is there a reason that we do not eliminate CryptoBinary from the XML Digital
SIgnature recommendation and only use base64Binary? Are the byte savings
significant in an XML context? Is CryptoBinary retained for backward
compatability with developing implementations, or is there another reason
I'm missing?

Thanks

---
Frederick Hirsch
Zolera Systems, http://www.zolera.com/
Information Integrity, XML Security