Re: Small patch to enable use of gpg-agent with pgg

[Simon Josefsson]

Sascha Wilde <address@hidden> writes:

>> Is there a choice of utilizing --status-fd output?  That is, query a
>> user for his passphrase only when gpg requested (i.e. we see
>> NEED_PASSPHRASE)
>
> I don't understand this point -- with the proposed patch the user
> already only gets queried for the passphrase when gpg needs it -- or
> am I missing something?

Right now, pgg assume that gpg will need a passphrase so it asks the
user for one.  It may be better if pgg postpone the passphrase query
until gpg tell pgg that it needs a passphrase.  Then it is only
queried for when it is needed.  This should be more reliable, but
slightly more complicated to implement (asynchronous code).

Gpg doesn't need a passphrase if the agent is used, or (I guess) if
the key isn't protected with a password.  If a smartcard is used, gpg
need a pin code and not a passphrase, I'm not sure pgg can handle this
at all now.  (This last point is separate, and wouldn't be solved if
we implement Daiki's idea.)

I may have misunderstood Daiki though...