On Tue, May 12, 2009 at 4:34 AM, Dan McGee wrote:

On Sun, May 10, 2009 at 6:16 AM, Nagy Gabor wrote:

...

...

From 1572535efe01dfae1f28b74862dec21f2102f165 Mon Sep 17 00:00:00 2001 From: Nagy Gabor Date: Sun, 10 May 2009 14:08:04 +0200 Subject: [PATCH] Don't print "local/" with -Qs

See FS#14642.

Signed-off-by: Nagy Gabor

Any objections to this? It is a longstanding "tradition" that we've had the local prefix, so I want to make sure this isn't a mistake, as this will definitely affect parsing of -Qs information as it can no longer be treated the same as -Ss information.

I expressed my thoughts in the bug report that we keep it as is; I'll await a few more opinions before taking this route. If we do take it, I expect someone to fix the pacsearch script in contrib. :)

It always bugged me that pacman outputs package names in a format that it cannot handle itself. If we want to keep the local/ prefix, we should also support it on every local operations. Not only -R, but also -Q, -Qi, etc.. I am curious about the operations on sync packages though, for example : does pacman -Si handle the repo/ format?

On Wed, May 13, 2009 at 2:50 PM, Dan McGee wrote:

Yes, definitely- all -S operations do so:

dmcgee@galway ~ $ pacman -Si testing/mlocate Repository     : testing Name           : mlocate Version        : 0.22-1 <snip>

dmcgee@galway ~ $ pacman -Si core/mlocate Repository     : core Name           : mlocate Version        : 0.21.1-1 <snip>

Ok good :) So to sum up : 1) if we keep local/ prefix, we need to fix all pacman operations dealing with local packages, not only -R 2) if we remove it, we need to fix tools like pacsearch