1 |
On Fri, 18 Nov 2005 19:09:57 -0800 Corey Shields <cshields@g.o> |
2 |
wrote: |
3 |
| I think having users systems would be profiled may help ease the |
4 |
| ricer issue. fex, user A has 3 systems, and marks package B as "!WFM" |
5 |
| on one. devs can cross link that negative mark to the system profile |
6 |
| and note that it's "-O12 --omg-itsofast", and disregard the negative |
7 |
| mark. You could even take it a step further and setup ratings for |
8 |
| the registered users, and those who end up with a set negativity |
9 |
| don't count or something (for the ricers).. |
10 |
|
11 |
The problem isn't so much people marking stuff as broken when it's not |
12 |
as people marking stuff as working when it isn't. Classic example: |
13 |
anything related to ricerfs or gcc-4. |
14 |
|
15 |
See, it's a question of quality rather than numbers. One "it works" |
16 |
report from someone who knows what they're doing is worth far more than |
17 |
a thousand "it works" reports from random users. Expecting a large |
18 |
number of average Joe types to produce useful testing reports is like |
19 |
expecting a large number of average Joe types to produce a Wikipedia |
20 |
article on how quantum cryptography works or a large number of average |
21 |
Joe types to produce a Gentoo Wiki article on the design and internal |
22 |
workings of versionator.eclass. |
23 |
|
24 |
| Just openly brainstorming here.. |
25 |
|
26 |
There was a similar proposal from (?)rac a couple of years back. Might |
27 |
be worth looking at why arch teams hated it last time around. |
28 |
|
29 |
-- |
30 |
Ciaran McCreesh : Gentoo Developer (Look! Shiny things!) |
31 |
Mail : ciaranm at gentoo.org |
32 |
Web : http://dev.gentoo.org/~ciaranm |