On Friday, July 8, 2011, Allan McRae firstname.lastname@example.org wrote:
On 08/07/11 23:55, Dan McGee wrote:
On Fri, Jul 8, 2011 at 6:59 AM, Allan McRaeallan@archlinux.org wrote:
The conversion to using parse_options causes this option to break. It is preferable to remove the option rather than fix it as it is simply a wrapper for "gpg --homedir @sysconfdir@/pacman.d/gnupg". Any user using more advanced keyring management than provided by pacman-key can manage to point gpg at the right place themselves...
How to manually edit the keyring with gpg will instead be documented in the man page in a later commit.
I won't lie here, I'm not a fan of this but maybe because I've become accustomed to the option being available. It was way easier than typing out the long-form gpg command line. "pacman-key --adv --verify /tmp/cryptsetup-1.3.1-1-x86_64.pkg.tar.xz.sig" is something I just pulled out of my command history.
What if we just enforced instead that the entire arg string was quoted: pacman-key --adv "--verify /tmp/cryptsetup-1.3.1-1-x86_64.pkg.tar.xz.sig" Or perhaps the "don't parse anymore" option: pacman-key --adv -- --verify /tmp/cryptsetup-1.3.1-1-x86_64.pkg.tar.xz.sig
The latter would work, but I am still not entirely convinced about the need for this...
I intend to add a --verify option to pacman-key because I think that would be a fairly common command to use. Anything else with common usage should also be added to pacman-key.
Is there anything else you used this for? I just have this nagging feeling that hiding what gpg is doing (we already have --no-permission-warning there by default) is not the way to go. Not that my opinion is overly strong on this.
That works too- if we add a --verify then I'm fine with this patch.
The --no-perm-warn is likely something we can/should move to a default gpg.conf file.