1 |
Daniel Goller wrote: |
2 |
|
3 |
>On Tue, 2006-04-11 at 09:36 -0400, Stephen P. Becker wrote: |
4 |
> |
5 |
> |
6 |
>>Eldad Zack wrote: |
7 |
>> |
8 |
>> |
9 |
>>>Hello, |
10 |
>>> |
11 |
>>>Sometimes it becomes a problem whenever a new release or a tricky bugfix comes |
12 |
>>>up for a certain package. |
13 |
>>>To improve QA we can let our userbase help, especially people who use certain |
14 |
>>>packages quite heavily - they can provide good or even superior QA than devs. |
15 |
>>> |
16 |
>>>I think it would be a nice idea to keep a userlist for anyone who'd like to |
17 |
>>>volunteer testing packages they regularly use. |
18 |
>>> |
19 |
>>>We can consider a web interface for enrolling users to specific packages, and |
20 |
>>>maybe even get a bug.g.o account for the list, this way a bug can be opened |
21 |
>>>for the testers to comment on whenever a change that requires testing or |
22 |
>>>maybe just aiding arch teams to stablize packages. |
23 |
>>> |
24 |
>>>Maybe this was already pitched but it has just occured to me. |
25 |
>>> |
26 |
>>>Comments? |
27 |
>>> |
28 |
>>> |
29 |
>>> |
30 |
>>> |
31 |
>>Isn't this why we already have the arch tester position as described by |
32 |
>>GLEP 41 (http://www.gentoo.org/proj/en/glep/glep-0041.html)? |
33 |
>>Furthermore, are you saying that users would enroll themselves via this |
34 |
>>hypothetical web interface, or that an arch team would do so for users |
35 |
>>who have proven themselves to be worthy? If the former, this would be a |
36 |
>>serious step back in terms of QA (think about sorting out all the crap |
37 |
>>reports from ricer overlay users with OMGFAST CFLAGS from the decent |
38 |
>>ones). If the latter, I think the arch tester position already covers |
39 |
>>this sort of thing. |
40 |
>> |
41 |
>>-Steve |
42 |
>> |
43 |
>> |
44 |
> |
45 |
>didn't he ask for people who know a particular application very well? |
46 |
>i think there is a big difference between agreeing to test one |
47 |
>particular package since they know it very well and want to make sure |
48 |
>noone breaks it vs. being a full AT with all the things they get asked |
49 |
>to test |
50 |
> |
51 |
> |
52 |
I understand the idea, and I like it. However inorder to really get this |
53 |
to work smoothly and be useful some type of user-feedback tool that |
54 |
would help report back the exact build environment with times dates and |
55 |
warnings & errors + user notes would really make this type of system shine. |
56 |
|
57 |
For example, package xyz-1.1 comes out, and user is on this hypothetical |
58 |
list and gets notified of it. |
59 |
package xyz-1.1 is ~arch (given), user decides they want to test it and |
60 |
they emerge it the usual way. |
61 |
After emerging version 1.1, user should (inorder to give his report) run |
62 |
a tool that will send the ebuild's environment (CFLAGS, etc) and prompt |
63 |
for the user to it a "rating" (value of whether or not the package |
64 |
works) & give some notes (say, special requirments to make it work, or a |
65 |
patch, or just simply , "Works".) |
66 |
|
67 |
Anyways, I like the idea. +1 |
68 |
|
69 |
Regards, |
70 |
Andrew |
71 |
-- |
72 |
gentoo-dev@g.o mailing list |