Different recipient match algorithms

[Stephan Mueller]

Hi,

When using gpg to encrypt to a recipient, the pattern matching logic to 
resolve a user-provided recipient identifier to a key seems to be performed 
with classify_user_id.

When applying a group definition expansion, the function expand_id performs 
the pattern matching by using strcasemp when searching for the group 
reference.

With this difference, the following is apparent: If I have 2 keys in my key 
ring with both looking like in that order:

AB <ab at c.d>
B <b at c.d>

both recipient resolution operations produce different results when encrypting 
to b at c.d. The group resolution resolves ab at c.d whereas the "regular" 
resolution operation returns b at c.d.

May I ask why this apparent different key resolution operation? Shouldn't 
expand_id also use classify_user_id instead of strcasecmp?

Ciao
Stephan