1 |
"Mark Haney" <mhaney@××××××××××××.org> posted |
2 |
4A72E4DC.2070106@××××××××××××.org, excerpted below, on Fri, 31 Jul 2009 |
3 |
08:34:36 -0400: |
4 |
|
5 |
|
6 |
> Here's my take on this, since I am OP. For the last year or two, I've |
7 |
> had, more and more, to go straight to ~arch for 'stable' packages. This |
8 |
> isn't so much about KDE4, which I /expected/ to be funky when it was |
9 |
> released, it's virtually everything else. Unlike some people, or most, |
10 |
> if I read the list right, are already running ~amd64 on their systems. |
11 |
> |
12 |
> I am not. |
13 |
> |
14 |
> And I do not want to. What's the point in having 'stable' when |
15 |
> virtually no packages are marked as such any more? I've been running |
16 |
> qt4.5 for nearly a year now. Isn't it about bloody time it gets marked |
17 |
> stable? Hell, IIRC, KDE3.5.10 isn't even marked stable (or wasn't last |
18 |
> time I looked). |
19 |
|
20 |
Note that while Gentoo does provide the tools to mix stable and ~arch for |
21 |
those who wish to try, it's not well tested or supported. All stable is |
22 |
supposed to be well tested, and all ~arch should be tested working at |
23 |
least on the Gentoo package maintainer's machines with all ~arch, but |
24 |
mixing the two is asking for trouble. |
25 |
|
26 |
Meanwhile, one of the issues with stable (which people like me parse as |
27 |
stale with a "b" added... hmm... I like that description, I think I'll |
28 |
keep using it! =:^) is that it takes people to test it that are willing |
29 |
to stay as far back as stale is in general, so they're testing a stale |
30 |
install not an ~arch install, but who are also willing to test candidates |
31 |
for stale, ensure they work, and report them as working. |
32 |
|
33 |
You say you have a big package.keywords file right now. But have you |
34 |
been reporting as working packages as you install them, so that the |
35 |
Gentoo/amd64 folks know they are working? (Maybe you are, I don't know, |
36 |
and I'm not saying you have to, but somebody has to, and those like me |
37 |
that want ~arch systems aren't going to be able to test with stale, and |
38 |
those who never install ~arch packages at all aren't going to be testing |
39 |
since they stick to stale, so the only ones left to test and report |
40 |
working are those who are mostly stale but do install the occasional not- |
41 |
yet-stale package. |
42 |
|
43 |
> I make the comment about it being right for me because I have been |
44 |
> getting the feeling Gentoo is becoming 'Debian v2.0' by just leaving |
45 |
> everything useful in ~arch (or testing in Debian's case). |
46 |
|
47 |
I've never run Debian, and prefer not to run stale anyway, so can't |
48 |
honestly say, there. But what I can say is that in all honesty, at least |
49 |
for me, if Gentoo/amd64 dropped stable all together (as is the case with |
50 |
a couple of the obscure archs, labeled 'experimental') it would only be |
51 |
beneficial to me, as that would be more testing and developer power for |
52 |
the ~arch I'm actually running. |
53 |
|
54 |
But I'm not selfish enough to wish that on the folks running stable. I'm |
55 |
just not interested in something that stale, is all, so it doesn't matter |
56 |
to me one way or the other. |
57 |
|
58 |
> If it is STABLE, mark it as such. Don't sit here and tell me, 'Oh just |
59 |
> run ~amd64 widget, it's stable'. |
60 |
|
61 |
They DO mark it as such, when they KNOW it's stale. But they don't KNOW |
62 |
it's stale, unless it has been reasonably tested on an otherwise stale |
63 |
system, and the people willing to do that testing and actually report the |
64 |
results aren't so easy to come by, so it's taking longer to know it's |
65 |
stale. |
66 |
|
67 |
> When I started with Gentoo in 2005/6, I could emerge -uD world and know |
68 |
> it'll pull in the latest stable packages and be done with it. Now, I |
69 |
> have to watch because some packages aren't, some might need a downgrade |
70 |
> of a package, which I have to mask so it doesn't get downgraded, ad |
71 |
> infinitum. |
72 |
|
73 |
Well, as I said, mixed stale and unstale isn't well tested or supported. |
74 |
But if it's marked stale and is removed, forcing a downgrade, there's a |
75 |
reason for it. OTOH if you package.keyworded as specific version, and it |
76 |
goes away, then yes, portage will suggest a downgrade. But that could |
77 |
only happen because it wasn't considered stale in the first place, and as |
78 |
an ~arch package with newer ones available, it can be removed. |
79 |
|
80 |
> To me, the distro is just feeling kinda sloppy on the back end. No, I'm |
81 |
> not looking for a 'Ubuntu' experience. That distro gives me heartburn. |
82 |
> But, geez, I do expect packages to be moved from testing to stable |
83 |
> slightly more often than never. I'm not trying to be overly critical |
84 |
> here, but the way things are going, it's getting /harder/ to maintain a |
85 |
> STABLE system now than it used to be. |
86 |
|
87 |
It may indeed be getting harder to maintain a stale system. I wouldn't |
88 |
know. But what I can say is what I said above, the only way the packages |
89 |
are going to move to stale is if people on stale test them and say they |
90 |
work on stale. |
91 |
|
92 |
> And, FWIW, on topic, having qt3support globally makes no difference. I |
93 |
> still have a thousand bleeding hoops to jump through to fix all the |
94 |
> asinine blocks and dependencies. |
95 |
|
96 |
It's really not that bad. As revealed in a different post, you had parts |
97 |
of and old qt4 blocking the newer version. As I said, trying to do |
98 |
multiple versions of the qt4 components doesn't work. It has to be all |
99 |
one version. Again as I said, once it's all one version, if you do |
100 |
--update --deep, the system should pretty much take care of its own |
101 |
updates. But it can't do that safely with split versions, and it'll have |
102 |
difficulty doing it if you don't run your updates with --deep, as well. |
103 |
|
104 |
So you're breaking at least one and possibly two rules (in addition to |
105 |
running an unsupported partially ~arch and partially stale arch system), |
106 |
trying to have multiple qt4 versions, and possibly, trying to update |
107 |
without using --deep, thus only making more trouble for yourself. |
108 |
|
109 |
If it's that far behind, why not run ~arch? A number of users have |
110 |
reported that full ~arch has actually been more stable for them than |
111 |
partially stale, partially ~arch. Either you want stale or you don't, |
112 |
and it doesn't appear you're satisfied with it, so... |
113 |
|
114 |
-- |
115 |
Duncan - List replies preferred. No HTML msgs. |
116 |
"Every nonfree program has a lord, a master -- |
117 |
and if you use the program, he is your master." Richard Stallman |