On Tue, Oct 18, 2016 at 09:19:22AM +0200, Landry Breuil wrote:
> So the original idea was more or less to use -z to 'fake' branch
> support when it wasnt existing.

Maby this is a good time to retire the local patch if the branch syntax
solves those problems for newer systems?

> As for pkg_info -e vs pkg_info -Iq, i dont know what was the intent of
> the change nor if it was good or not

The reason for the change was to make the branch syntax work. -Iq
inst:<name> can expand a name based on the branch syntax to the actual
installed version, which -e can not do:
$ pkg_info -e autoconf%2.13
Invalid spec: autoconf%2.13

$ pkg_info -Iq inst:autoconf%2.13

> but here -Iq finds an installed
> package by giving only its name (but not if you pass the version without
> patchlevel):
> $pkg_info -Iq ansible
> ansible-
> $pkg_info -e ansible
> Invalid spec: ansible
> $pkg_info -e ansible-
> inst:ansible-
> $pkg_info -Dunsigned -Iq ansible-
> Error from 
> http://ftp.fr.openbsd.org/pub/OpenBSD/snapshots/packages/amd64/ansible-
> ftp: Error retrieving file: 404 Not Found
> $pkg_info -Iq ansible-
> ansible-
> The only 'advantage' of -e here is that it finds it if you pass the full
> version without the patchlevel.

Right, the reason for the switch was the requirement to find the real
package on disk when a user supplies a name using the branch syntax.

Patrik Lundin

Reply via email to