[Duplicity-talk] strategy for error: sign+encrypt failed: unusable public key

[edgar . soldin]

Hello all,

somebody switching from an old to the recent  ftplicity version came up 
with an error similar to this...

gpg: FFFFFFFF: There is no assurance this key belongs to the named user
gpg: [stdin]: sign+encrypt failed: unusable public key
gpg: encrypted with 2048-bit ELG-E key, ID FFFFFFFF, created 2007-12-17
     "duplicity"
gpg: FFFFFFFF: There is no assurance this key belongs to the named user
gpg: [stdin]: sign+encrypt failed: unusable public key

this was because the selected key was not trusted, he didn't know why it 
suddenly happend, because the former combination of ftplicity 
1.1.1/duplicity 0.4.9/gpg-1.4.5-24.4 worked fine.... but still this can 
happen when installing/switching machines or accounts - so it should be 
prevented .. especially as the gpg error message only comes up with 
verbosity 5 or more

... this made me think of ways to prevent this error .. 
as I didn't find a way to let gpg show the trust state of a key, the 
only way for now is to test-sign+encrypt something and to check if that 
throws an error
e.g. > echo "$PASS" | gpg --passphrase-fd 0 -e -r FFFFFFFF --batch -o 
/tmp/mktemp.file test.txt

the other solution I think of is a bit more straight forward ... why not 
setting gpg --trust-always  .. as  the user selects  a key that he/her 
obviously wants to use and therefore has to trust
I am interested in opinions about this idea .. as there is currently no 
scenario I can imagine, except of a hacked backup user account (but then 
everything is lost already, so it doesn't matter),  where the 
--trust-always could be security problematic

on the other hand .. if there is an easy way to doublecheck if a key is 
trusted ultimately, I still would think about this way, as test 
encrypting someting does not seem very elegant to me


thanks for your thoughts .. ede


--
Edgar Soldin