[usmt dev]Another conundrum

[Mike Schiraldi]

So, how exactly should we deal with duplicate removal (avoidance)?

Let's say there's a key in the cache that reads:

Addr: foo@bar.com
Type: x509-cert
Data: sdaasdfkjasdf
Something-Else: foo

For each of the following, do we ignore it, merge it, or add it as a new
one?

Addr: foo@bar.com
Type: x509-cert
Data: sdaasdfkjasdf

Addr: foo@bar.com
Type: x509-cert
Something-Else: foo

Addr: foo@bar.com
Type: x509-cert
Data: sdaasdfkjasdf
Somethine-Else: bar

Addr: foo@bar.com
Type: x509-cert
Data: sdaasdfkjasdf
Something-Else: foo
Another-Key: blah

We could add a Unique-ID field as Thomas suggested, but that Unique-ID has
to be based on something. Perhaps the MD5 sum of <Addr, Type, Data>. Addr
and Type are already required; this would make Data required as well. 

But is it conceivable that someone one day might want two records with the
same Addr, Type, and Data? This approach would make that impossible.