[aur-general] TU Meeting

[Ryan Coyner]

On Sat, Nov 29, 2008 at 2:56 PM, w9ya <w9ya at qrparci.net> wrote:

>
>
> On Sat, Nov 29, 2008 at 12:26 PM, Loui Chang <louipc.ist at gmail.com> wrote:
>
>> I'll attempt to answer some of your questions.
>>
>> On Sat, Nov 29, 2008 at 11:39:44AM -0700, w9ya wrote:
>> > A few points of rebuttal. Mostly too many unanswered questions.
>> >
>> > I will outline these below, but for those of you that are becoming tired
>> of
>> > all this back and forth I can say this much:
>>
>> Bob, first of all please stop posting. If you're unable to properly
>> quote and address proper portions of someone's proposal then you're not
>> helping the discussion at all. It seems you're just trying to disorient
>> and befuddle the debate.
>
>
> Here i will middle post as you did. (Sometimes other forms of posting can
> make sense btw. So PLEASE do not put me onthe defensive.)
>

If I am not mistaken the general idea is to follow the natural flow of text,
which is top to bottom. Middle posting to comment a specific quote is fine
because it still preserves the flow of conversation when there are multiple,
nested quotes. If you top post this flow is broken, and it becomes more
difficult to follow a conversation.


>
>
>>
>> > 4 - There is absolutely no reason for this back and forth on simple
>> > questions like "show us your math. Please explain your assumptions.
>> (Both
>> > basic methodology questions.) Why are we being asked to do this now ?
>> Can
>> > you be more specific ?
>>
>> There is some data that Daenyth and I were collecting to see
>> how effective some of the proposed changes would be.
>>
>> [1] http://omploader.org/veDEy
>
>
> I cannot even begin to understand what this list of numbers means. Again,
> without simple and SPECFIC understandings, this is not of any value. AND it
> is important with statistics to not only understand (label) the collums you
> produce, but also what assumptions AND methodology were taken in their
> compilation.
>
> So far NO ONE has offered this up, so the numebrs have no USEFUl meaning.
>

This you blew out of proportion. I'm not even a TU, and by simply glancing
at it I can tell that it means:

[# of votes] [size of binary in kb] [name of package]

This becomes very clear if you look at the very bottom of the page. I do
agree that having more detailed reports would be nice, but that can take a
lot of time. I would say that the information Daenyth and Louipc collected
is good enough, and certainly better than nothing.


>
>
> Again I ask Why are we voting on numbers that onlyh one person has any
> remote understanding of ?
>
>>  <http://omploader.org/veDEy>
>>
>> > 5 - Since it has been declared that these are the first of many changes
>> to
>> > improve our resource utilization, what else is being planned or
>> considered ?
>>
>> If you've been following the news, there have been server upgrades.
>> I've also made a few changes in the community scripts that help slightly
>> and I plan to do even more. The last part of a total solution would be
>> to implement a minimum vote requirement, or suggestion. If not that,
>> then some way to minimise the number of packages that are largely
>> unused.
>
>
> Yes and I am happy that this has been taking place. I have offered money
> and no one has seen fit to take it as my part towards a solution for
> resource issues.
>
> But I ask yet again; How is removing 100 or so binary packages that you all
> are saying no one uses going to change anything for morew than a few weeks ?
>
> How are these "unused packages" representing a drain on resources ?
>
> How are releasing 2 % of storage space going to seriously and for the long
> term going to represent anything of value ?
>
> I can go on and on, as these questions are still not answered and are
> basic. It is quite simple to come up with them.
>

You guys have two separate problems.

1) The server is running low on resources.
2) The process of adding binary packages into the repository is not optimal.

The operative word is "separate." Problem #2 has a direct impact on Problem
#1, but they are still separate issues. Both of them needs to be addressed.



>
>
>>
>>
>> > 9 - Voting can be faked. How is this being dealt with ?
>>
>> Pkgstats can help verify the votes.
>
>
> Maybe, but not until they have been in place, **vetted**, and have been
> running for MANY MANY months.
>

You're never going to get a perfect system. Voting is flawed, but so is
pkgstats. Just use both. Together they can paint a more accurate picture
instead of relying on just one. If you can think of another way to measure
package use, throw that in the equation too.


-- 
Ryan Coyner
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://archlinux.org/pipermail/aur-general/attachments/20081129/c1505b31/attachment.htm>