[Bug] lingering gpg-agents from test suite

[Bug] lingering gpg-agents from test suite

[David Bremner]

Thanks to changes in gnupg 2.1.20, the agent started by the notmuch test
suite is no longer killed by deleting the home directory.

So yay, we get to adapt to changes in gnupg, again.

See also

    https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=862093

Re: [Bug] lingering gpg-agents from test suite

[Justus Winter]

[-- Attachment #1: Type: text/plain, Size: 1170 bytes --]

David Bremner <david@tethera.net> writes:

> Thanks to changes in gnupg 2.1.20, the agent started by the notmuch test
> suite is no longer killed by deleting the home directory.
>
> So yay, we get to adapt to changes in gnupg, again.
>
> See also
>
>     https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=862093

where you write:
> As of 2.1.20, removing $GNUPGHOME is no longer enough for gpg-agent to
> shut itself down. It seems that removing the directory
> /run/user/gnupg/d.$hash is enough to trigger the shutdown.
>
> I'm not sure the underlying cause; perhaps the change to use /run/user
> for non-standard homedirs (which I generally think is sensible) has
> this as a side-effect.

What triggers the automatic shutdown is the removal of the socket.
2.1.20 creates the so-called socket directory automatically (if it finds
/**/run/user).

> The main practical difficulty for me is that this makes it harder for
> test suites using gnupg to work across multiple gpg versions and
> multiple environments. I suppose I'll have to see how far back
> "gpgconf --kill all" is supported.

Yes, please try to do 'gpgconf --kill all' and ignore the exit status.


Justus

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

Re: [Bug] lingering gpg-agents from test suite

[Tomi Ollila]

On Mon, May 15 2017, Justus Winter wrote:

> David Bremner <david@tethera.net> writes:
>
>> Thanks to changes in gnupg 2.1.20, the agent started by the notmuch test
>> suite is no longer killed by deleting the home directory.
>>
>> So yay, we get to adapt to changes in gnupg, again.
>>
>> See also
>>
>>     https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=862093
>
> where you write:
>> As of 2.1.20, removing $GNUPGHOME is no longer enough for gpg-agent to
>> shut itself down. It seems that removing the directory
>> /run/user/gnupg/d.$hash is enough to trigger the shutdown.
>>
>> I'm not sure the underlying cause; perhaps the change to use /run/user
>> for non-standard homedirs (which I generally think is sensible) has
>> this as a side-effect.
>
> What triggers the automatic shutdown is the removal of the socket.
> 2.1.20 creates the so-called socket directory automatically (if it finds
> /**/run/user).
>
>> The main practical difficulty for me is that this makes it harder for
>> test suites using gnupg to work across multiple gpg versions and
>> multiple environments. I suppose I'll have to see how far back
>> "gpgconf --kill all" is supported.
>
> Yes, please try to do 'gpgconf --kill all' and ignore the exit status.

Quick & dirty:

trap 'if command -v gpgconf >/dev/null; then
         gpgconf --kill all || :
      fi
      trap - 0' 0 INT HUP TERM QUIT

I'll look a bit more a bit later...

Tomi


>
>
> Justus
> _______________________________________________
> notmuch mailing list
> notmuch@notmuchmail.org
> https://notmuchmail.org/mailman/listinfo/notmuch