1 |
On Wed, 2004-07-28 at 10:23 -0500, Caleb Tennis wrote: |
2 |
> On Wednesday 28 July 2004 09:41 am, Patrick Lauer wrote: |
3 |
> > On Wed, 2004-07-28 at 15:53, Caleb Tennis wrote: |
4 |
> > > I didn't intend for it to be package.mask'd - it's quite stable, |
5 |
> > |
6 |
> > I'd like to disagree: Bug 58527 |
7 |
> > Konqueror is consistently segfaulting on my machine, so I'd like to keep |
8 |
> > 3.3_beta masked until 3.3 final arrives. |
9 |
> > 3.3 is still a beta, so pushing it in ~x86 seems like a very strange |
10 |
> > decision to me. ~x86 is for unstable _ebuilds_, not unstable packages as |
11 |
> > far as I know. |
12 |
> |
13 |
> Because it consistently segfaults on your machine is no reason to keep it |
14 |
> masked - in fact, it works quite well for me and a large number of users or I |
15 |
> never would have considered it. Quite likely something else is going on. A |
16 |
> backtrace with debugging info would be helpful to solve it. |
17 |
|
18 |
Actually it is a good reason to keep it masked, especially because we're |
19 |
talking about a version here that is not by the upstream devs considered |
20 |
stable. The 'large number of users' point is moot, you cannot just say |
21 |
it works for 90% of the users & that makes it stable enough. If it |
22 |
doesn't work for 10% thats still substantial breakage (just making up |
23 |
numbers here). |
24 |
|
25 |
> > What's the policy on this? |
26 |
> > As much as I like bleeding edge stuff, I'd like to keep the bleeding as |
27 |
> > small as possible :-) |
28 |
> |
29 |
> It's far from bleeding edge - the features are frozen and it's namely just bug |
30 |
> fixes at this point. If anyone disagrees simply because the package name |
31 |
> contains the word "beta", then please come at me with more hard evidence. |
32 |
> |
33 |
> I assume that vapier thought I "forgot" to package.mask it, so he did it for |
34 |
> me, and this is why I unmasked it without throwing too much of a fuss. |
35 |
> |
36 |
> So, let's be productive here and use the time between now and the final |
37 |
> release to get the ebuilds and dependencies fixed, make sure things work |
38 |
> well, and hammer out bugs. If you still feel like it's too unstable to use, |
39 |
> then simply don't use it. :) |
40 |
|
41 |
~arch is for testing ebuilds, not packages. It's quite something that |
42 |
you decide something to be 'arch stable' before it is deemed stable by |
43 |
the upstream authors, who have a way more intimate knowledge of their |
44 |
code than you do. Plus that some arches have a liberal view on marking |
45 |
packages stable, so they might even make this beta stuff stable at some |
46 |
point. |
47 |
|
48 |
Anyway, how would you now test fixes for the 3.2 tree... the 3.3 ~arch |
49 |
beta packages get now chosen over ~arch 3.2 stuff ? |
50 |
|
51 |
I know KDE is all slotted and so on (which serves little purpose in |
52 |
itself besides usually taking up even more diskspace on user machines) |
53 |
and so it doesn't have to be a big problem, but in general i consider it |
54 |
very bad behaviour QA wise to make beta stuff available to the general |
55 |
public (and that is what happens in ~arch). |
56 |
|
57 |
- foser |