Dan McGee wrote:

On Dec 28, 2007 7:16 PM, Scott Horowitz wrote:

...

On Dec 28, 2007 2:19 PM, wrote:

...

This is a patch to make Pacman only ask for confirmation if anything different from what the user requested (i.e. dependencies) is going to be installed. Since the user took the time to type out "pacman -Sy foo", they obviously did want foo installed, and it really wouldn't make sense to ask them this again unless something else will be installed.

I like this. It's akin to the fact that pacman -R foo doesn't prompt you.

(How crazy is that? Removing a package doesn't prompt you but installing a package does.)

-R never resolves dependencies. -A/-U never resolve dependencies. -S *does* resolve dependencies. Thus the difference.

I like the idea, although I may disagree that people expect it to be installed immediately instead of getting confirmation. Did you test this in the replaces case? Something like pkgA replaces pkgB, you have pkgB installed, and you run 'pacman -S pkgA'. I'd be caught awfully off guard if pacman just went ahead in this case, but the target list wouldn't grow so I'm guessing your patch would have unintended consequences here.

Thoughts from the rest of the crew?

-Dan

_______________________________________________ pacman-dev mailing list pacman-dev@archlinux.org http://archlinux.org/mailman/listinfo/pacman-dev

Good point - I didn't think about that. I just quickly read through some of the pacman code before doing this - I still don't completely understand it. If there's an alpm_list_compare or something like that, that would be better. But that's a job for someone else, as I'd probably break something :-) (Just a note: I tested this with a package with no deps, 2 packages with no deps, a package with deps and a package without, and a package with deps. I didn't even think of replacements.)

On Sat, Dec 29, 2007 at 1:36 AM, Aaron Griffin wrote:

On Dec 28, 2007 7:33 PM, Dan McGee wrote:

...

On Dec 28, 2007 7:16 PM, Scott Horowitz wrote:

...

On Dec 28, 2007 2:19 PM, wrote:

...

This is a patch to make Pacman only ask for confirmation if anything different from what the user requested (i.e. dependencies) is going to be installed. Since the user took the time to type out "pacman -Sy foo", they obviously did want foo installed, and it really wouldn't make sense to ask them this again unless something else will be installed.

I like this. It's akin to the fact that pacman -R foo doesn't prompt you.

(How crazy is that? Removing a package doesn't prompt you but installing a package does.)

-R never resolves dependencies. -A/-U never resolve dependencies. -S *does* resolve dependencies. Thus the difference.

I like the idea, although I may disagree that people expect it to be installed immediately instead of getting confirmation. Did you test this in the replaces case? Something like pkgA replaces pkgB, you have pkgB installed, and you run 'pacman -S pkgA'. I'd be caught awfully off guard if pacman just went ahead in this case, but the target list wouldn't grow so I'm guessing your patch would have unintended consequences here.

Thoughts from the rest of the crew?

I'm with Dan here. As long as it's only doing what's requested (i.e. not replacing anything), then it's kosher.

Ping? It was email cleanup night here and I came across this. -Dan

On Mon, Mar 10, 2008 at 7:56 PM, Dan McGee wrote:

On Sat, Dec 29, 2007 at 1:36 AM, Aaron Griffin wrote:

...

On Dec 28, 2007 7:33 PM, Dan McGee wrote:

...

On Dec 28, 2007 7:16 PM, Scott Horowitz wrote:

...

On Dec 28, 2007 2:19 PM, wrote:

...

This is a patch to make Pacman only ask for confirmation if anything different from what the user requested (i.e. dependencies) is going to be installed. Since the user took the time to type out "pacman -Sy foo", they obviously did want foo installed, and it really wouldn't make sense to ask them this again unless something else will be installed.

I like this. It's akin to the fact that pacman -R foo doesn't prompt you.

(How crazy is that? Removing a package doesn't prompt you but installing a package does.)

-R never resolves dependencies. -A/-U never resolve dependencies. -S *does* resolve dependencies. Thus the difference.

I like the idea, although I may disagree that people expect it to be installed immediately instead of getting confirmation. Did you test this in the replaces case? Something like pkgA replaces pkgB, you have pkgB installed, and you run 'pacman -S pkgA'. I'd be caught awfully off guard if pacman just went ahead in this case, but the target list wouldn't grow so I'm guessing your patch would have unintended consequences here.

Thoughts from the rest of the crew?

I'm with Dan here. As long as it's only doing what's requested (i.e. not replacing anything), then it's kosher.

Ping? It was email cleanup night here and I came across this.

I think my original sentiment still stands. apt-get does the same thing: if I am installing 'pkgfoo' and there are no deps, and only pkgfoo is to be installed, it skips the yes/no step.

2008/3/12, Nagy Gabor :

...

I think my original sentiment still stands. apt-get does the same thing: if I am installing 'pkgfoo' and there are no deps, and only pkgfoo is to be installed, it skips the yes/no step.

Well, I don't know. This would be a bit disturbing to me (sometimes I must confirm, sometimes I needn't). And I like the current behavior: I check the download size etc. then I confirm.

I've used to the same behaviour. -- Roman Kyrylych (Роман Кирилич)