On Wed, Dec 3, 2008 at 3:40 PM, w9ya <w9ya@qrparci.net> wrote:
> On Wed, Dec 3, 2008 at 2:23 PM, Aaron Griffin <aaronmgriffin@gmail.com>
> wrote:
>>
>> On Wed, Dec 3, 2008 at 3:15 PM, Kristoffer Fossgård <kfs1@online.no>
>> wrote:
>> > Why is package popularity judged by votes anyway? I never vote. The
>> > reason i never vote is because i don't understand why package
>> > popularity can't simply be judged by download count. I know there's
>> > been some discussion of this topic earlier but as far as i can recall
>> > there were no convincing arguments against it. even if some moron
>> > decides to download his package more times to increase the chance of
>> > adoption(where would the motivation to do this be anyway? if he simply
>> > want's to get it in the pacman system for easier maintenance a simple
>> > guide to make your own repository and add it to pacman would remove
>> > this incurrence in 99% of cases) the TU's and Devs could still choose
>> > to not include the package in the repos. This could also largely be
>> > avoided by only counting i guess certain ip ranges(i'm not an expert on
>> > these things, but i DO know that counting downloads with some level of
>> > security is a common occurence on the net)
>>
>> We have mirrors. Almost 100 of them. Feel free to contact them all,
>> have them write code to count downloads which then sends the stats to
>> us, and then we can implement this.
>>
>> What you suggest is absolutely not feasible at all.
>
> Quite frankly Aaron, this attitude is not helpful to your case at all. And
> it leads to worse as we have seen between you and me as of late. No one
> likes to be rebutted in such a manner.
>
> Yes, it is NOT feasible, BUT you can **choose** to say this nicely or
> coarsely.

Kristoffer, I apologize if this sounded harsh, as Mr Finch seemed to
interpret it. I did not mean it as such - I meant to say that you were
overlooking the fact that we do not have full control over our mirrors
and can only track downloads from one out of *many* servers.

On Wed, Dec 3, 2008 at 2:58 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:

On Wed, Dec 3, 2008 at 3:40 PM, w9ya <w9ya@qrparci.net> wrote:

> On Wed, Dec 3, 2008 at 2:23 PM, Aaron Griffin <aaronmgriffin@gmail.com>

> wrote:

>>

>> On Wed, Dec 3, 2008 at 3:15 PM, Kristoffer Fossgård <kfs1@online.no>

>> wrote:

>> > Why is package popularity judged by votes anyway? I never vote. The

>> > reason i never vote is because i don't understand why package

>> > popularity can't simply be judged by download count. I know there's

>> > been some discussion of this topic earlier but as far as i can recall

>> > there were no convincing arguments against it. even if some moron

>> > decides to download his package more times to increase the chance of

>> > adoption(where would the motivation to do this be anyway? if he simply

>> > want's to get it in the pacman system for easier maintenance a simple

>> > guide to make your own repository and add it to pacman would remove

>> > this incurrence in 99% of cases) the TU's and Devs could still choose

>> > to not include the package in the repos. This could also largely be

>> > avoided by only counting i guess certain ip ranges(i'm not an expert on

>> > these things, but i DO know that counting downloads with some level of

>> > security is a common occurence on the net)

>>

>> We have mirrors. Almost 100 of them. Feel free to contact them all,

>> have them write code to count downloads which then sends the stats to

>> us, and then we can implement this.

>>

>> What you suggest is absolutely not feasible at all.

>

> Quite frankly Aaron, this attitude is not helpful to your case at all. And

> it leads to worse as we have seen between you and me as of late. No one

> likes to be rebutted in such a manner.

>

> Yes, it is NOT feasible, BUT you can **choose** to say this nicely or

> coarsely.



Kristoffer, I apologize if this sounded harsh, as Mr Finch seemed to

interpret it. I did not mean it as such - I meant to say that you were

overlooking the fact that we do not have full control over our mirrors

and can only track downloads from one out of *many* servers.


A better attempt. Thank you.

Bob F.

On Wed, 3 Dec 2008 15:58:03 -0600, "Aaron Griffin"
<aaronmgriffin@gmail.com> wrote:
> On Wed, Dec 3, 2008 at 3:40 PM, w9ya <w9ya@qrparci.net> wrote:
>> On Wed, Dec 3, 2008 at 2:23 PM, Aaron Griffin <aaronmgriffin@gmail.com>
>> wrote:
>>>
>>> On Wed, Dec 3, 2008 at 3:15 PM, Kristoffer Fossgård <kfs1@online.no>
>>> wrote:
>>> > Why is package popularity judged by votes anyway? I never vote. The
>>> > reason i never vote is because i don't understand why package
>>> > popularity can't simply be judged by download count. I know there's
>>> > been some discussion of this topic earlier but as far as i can recall
>>> > there were no convincing arguments against it. even if some moron
>>> > decides to download his package more times to increase the chance of
>>> > adoption(where would the motivation to do this be anyway? if he
> simply
>>> > want's to get it in the pacman system for easier maintenance a simple
>>> > guide to make your own repository and add it to pacman would remove
>>> > this incurrence in 99% of cases) the TU's and Devs could still choose
>>> > to not include the package in the repos. This could also largely be
>>> > avoided by only counting i guess certain ip ranges(i'm not an expert
> on
>>> > these things, but i DO know that counting downloads with some level
> of
>>> > security is a common occurence on the net)
>>>
>>> We have mirrors. Almost 100 of them. Feel free to contact them all,
>>> have them write code to count downloads which then sends the stats to
>>> us, and then we can implement this.
>>>
>>> What you suggest is absolutely not feasible at all.
>>
>> Quite frankly Aaron, this attitude is not helpful to your case at all.
> And
>> it leads to worse as we have seen between you and me as of late. No one
>> likes to be rebutted in such a manner.
>>
>> Yes, it is NOT feasible, BUT you can **choose** to say this nicely or
>> coarsely.
>
> Kristoffer, I apologize if this sounded harsh, as Mr Finch seemed to
> interpret it. I did not mean it as such - I meant to say that you were
> overlooking the fact that we do not have full control over our mirrors
> and can only track downloads from one out of *many* servers.
>

Not that I want to jump right in the middle, but I agree with Aaron. I
didn't take his prev email as harsh and if you want to go down that road
Bob your could be interpeted wrongly also.

> P.S... I am *still* waiting for answers to my questions from yesterday.
But please take as much time as you need to answer them.

That could be interpreted as being a smart ass.

I really am not calling you a smart ass Bob. I have no problems with you I
just want to point out that I think that people on both sides are getting a
little sensetive with this. I see good points from both sides of the fence.

Just an outside observation.

Brandon
--
Brandon Martin

On Wed, Dec 3, 2008 at 3:57 PM, Loui Chang <louipc.ist@gmail.com> wrote:
> On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
>> Why is package popularity judged by votes anyway? I never vote. The
>> reason i never vote is because i don't understand why package
>> popularity can't simply be judged by download count.
>
> Votes are a conscious approval of a package. Downloads are not.

Interesting distinction I haven't heard spelled out in words: explicit
approval vs implicit approval. 8)

Loui Chang wrote:
> On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
>> Why is package popularity judged by votes anyway? I never vote. The
>> reason i never vote is because i don't understand why package
>> popularity can't simply be judged by download count.
>
> Votes are a conscious approval of a package. Downloads are not.
>

explicit downloads sure are... unless you like to play pacman roulette,
i guess.

but i don't. if it's explicitly installed, it's because i wanted the
damn thing.

-kludge

kludge wrote:
> Loui Chang wrote:
>> On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
>>> Why is package popularity judged by votes anyway? I never vote. The
>>> reason i never vote is because i don't understand why package
>>> popularity can't simply be judged by download count.
>> Votes are a conscious approval of a package. Downloads are not.
>>
>
> explicit downloads sure are... unless you like to play pacman roulette,
> i guess.
>
> but i don't. if it's explicitly installed, it's because i wanted the
> damn thing.
>
> -kludge

sorry, that obviously only applies to [community] and to the aur. but
it still applies to [community].

-kludge

On Wed, Dec 3, 2008 at 7:21 PM, kludge <drkludge@rat-patrol.org> wrote:
> Loui Chang wrote:
>> On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
>>> Why is package popularity judged by votes anyway? I never vote. The
>>> reason i never vote is because i don't understand why package
>>> popularity can't simply be judged by download count.
>>
>> Votes are a conscious approval of a package. Downloads are not.
>>
>
> explicit downloads sure are... unless you like to play pacman roulette,
> i guess.
>
> but i don't. if it's explicitly installed, it's because i wanted the
> damn thing.

And, again, tracking downloads is not feasible. I mean, technically we
can't even track downloads from ftp.archlinux.org, as even THAT is a
mirror

On Wed, Dec 03, 2008 at 07:21:32PM -0600, kludge wrote:
> Loui Chang wrote:
> > On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
> >> Why is package popularity judged by votes anyway? I never vote. The
> >> reason i never vote is because i don't understand why package
> >> popularity can't simply be judged by download count.
> >
> > Votes are a conscious approval of a package. Downloads are not.
> >
>
> explicit downloads sure are... unless you like to play pacman roulette,
> i guess.
>
> but i don't. if it's explicitly installed, it's because i wanted the
> damn thing.

That's good. I occaisionally download apps to try them and remove them
soon after. Those are packages I would not vote on.

Don't tell me you haven't ever done that.

On Wed, Dec 03, 2008 at 08:31:05PM -0600, kludge wrote:
> >> but i don't. if it's explicitly installed, it's because i wanted the
> >> damn thing.
> >
> > That's good. I occaisionally download apps to try them and remove them
> > soon after. Those are packages I would not vote on.
> >
> > Don't tell me you haven't ever done that.
>
> fair enough.
>
> what about votes:downloads, accompanied by a big push to get people voting?

This proposal should be the signal that we are going to start taking
votes more seriously, but no packages will be removed for now.

> (btw, as a result of this debate, i'm finally voting for the packages i
> use.)

Good to hear!

Here's a little script to help figure out what's installed from a
particular repo. Use `repo-installed community` for [community], etc.
This'll help people figure out what to vote for.

#!/bin/bash
# Print a list of packages installed from a given repo.
#
# Not everything will be listed if your installed packages
# are out of date.
#
# This will work for most standard installations.
# If you have an exotic setup just edit pacmandb.
#
# Usage: repo-installed <repo>

pacmandb=/var/lib/pacman

if [ -z "$1" ] ; then
echo "Please specify a repo."
echo -e " Usage: $0 <repo>"
exit 1
fi

installed=($(/bin/ls "$pacmandb/local"))

for pkg in ${installed[@]}
do
if [ -e "$pacmandb/sync/$1/$pkg" ] ; then
echo $pkg
fi
done

>On Wed, Dec 03, 2008 at 10:15:23PM +0100, Kristoffer Fossgård wrote:
>> Why is package popularity judged by votes anyway? I never vote. The
>> reason i never vote is because i don't understand why package
>> popularity can't simply be judged by download count.
>
>Votes are a conscious approval of a package. Downloads are not.
>
Your all missing my point. I never said counting packages by
downloadrate is a perfect solution but that IT IS GOOD ENOUGH _and_
BETTER THAN THE VOTE SYSTEM.

Think about it, the times a user downloads a package because he WANTS
the package far superceeds the times a user downloads a package because
he might just want to check it out, this goes for every user, thus
making the system work.

And what goes for "conscious" vs "unconscious"; if i get another
package as a dep to a package i most likely WANT, I _want_ it. Even
though i just want to test it out that is *interest*!(and i can't
rememer last time i did this..)
>We have mirrors. Almost 100 of them. Feel free to contact them all,
>have them write code to count downloads which then sends the stats to
>us, and then we can implement this.

Put the counting algorithm directly into pacman then and then do
the same for aur. Problem solved.

The alternative here is putting another burden on users which they
surely will not follow up on(thus making the system not work anyway) and
also burdening the devs and TU's with unneccesary policy they don't
need.

-K