1 |
On Fri, May 9, 2014 at 8:28 PM, Patrick Lauer <patrick@g.o> wrote: |
2 |
> THAT is the bottleneck - getting people motivated to triage logs (which can be |
3 |
> automated quite far) and file bugs (which is demotivating and exhausting). You |
4 |
> fix that problem and the rest is easy ... |
5 |
|
6 |
Agree for the most part. |
7 |
|
8 |
Also, nobody needs "permission" to run a tinderbox. If you want to |
9 |
run one, just do it! Complaining about other people not doing it |
10 |
isn't terribly productive. |
11 |
|
12 |
The typical tinderbox approach I've seen is that you have a suite of |
13 |
tests that get run as frequently as the box can manage, and if a test |
14 |
fails everybody stops committing and cleans up the tree. Such a model |
15 |
basically makes dealing with errors everybody's problem, and not just |
16 |
the tinderbox guy's. |
17 |
|
18 |
The problem is that with something like Gentoo the permutations are |
19 |
fairly extensive, the tree is extremely diverse, and I could see |
20 |
countless arguments about what constitutes an error that needs to be |
21 |
fixed. That might be part of why nobody wants to run a tinderbox. |
22 |
|
23 |
If you want to find bugs, though, a real easy way to start is to just |
24 |
unpack the current stage3, set the profile to kde or gnome, stick |
25 |
gnome/kde-meta in world, and do an emerge -auDNv world. There is a |
26 |
decent chance that you'll run into a build failure, just like anybody |
27 |
trying to install Gentoo for the first time. Seems like a good place |
28 |
to start if anybody wants to do a tinderbox (note that I haven't done |
29 |
this too recently - for a while I was doing it monthly and filing bugs |
30 |
fairly consistently)... |
31 |
|
32 |
Rich |