1 |
On Tue, 20 Jan 2009 21:36:58 +0100 |
2 |
"b.n." <brullonulla@×××××.com> wrote: |
3 |
|
4 |
> Mark Knecht ha scritto: |
5 |
> |
6 |
> > The one thing I would respectfully suggest is that you carefully |
7 |
> > build your own portage overlay. My experience with Gentoo over the |
8 |
> > last few years is that there is a _anxiousness_ in the portage |
9 |
> > maintainer area to move newer revisions of software into portage |
10 |
> > quickly and then just as quickly to remove from portage what users |
11 |
> > are currently using. |
12 |
|
13 |
@Mark |
14 |
That's certainly true in the sense that we loathe maintaining |
15 |
several revisions of the same software. Each Gentoo maintainer can |
16 |
maintain anywhere from 1 to $BIG_NUM packages, so we strive to have in |
17 |
general at most three versions in portage at any given time. We don't |
18 |
really want bug reports about $old_stable if it's been fixed in a |
19 |
$new_stable. We're not backport-monkeys, like Ubuntu. We do what we do |
20 |
cause we like solving complex problems, interacting with the smart |
21 |
people we call 'users'[1] and our fellow devs, not because "svn diff" |
22 |
is our BFF. :-) |
23 |
|
24 |
> I am usually a bit annoyed by the contrary. On an almost 1-year old |
25 |
> Kubuntu (8.04 Hardy Heron) I can find packages that are just barely |
26 |
> x86 stable now on Gentoo. |
27 |
> |
28 |
> A couple of examples I am aware of: |
29 |
> Firefox 3: stable just since one month on Gentoo x86, was included in |
30 |
> KB8.04 Qtiplot: 0.9.x stable and working on KB8.04, all releases ~x86 |
31 |
> (and a hell to compile on a stable system -still didn't manage to do |
32 |
> it) in Gentoo. |
33 |
|
34 |
I don't know about qtiplot but Firefox-3 was blocked by the fact that |
35 |
there were stability problems the first many months, compared to |
36 |
firefox-2. I remember random crashes, etc. Then we had a mysterious bug |
37 |
where it would segfault on first start if compiled with |
38 |
USE="xulrunner", i.e. using the system libxul, but not if we used the |
39 |
bundled one. Then we had some problems with hardened Gentoo, Sparc |
40 |
getting bus errors, etc. If you remember firefox-2 when first it came |
41 |
out, it also had the same kinds of problems. I think it wasn't before |
42 |
2.0.0.11 that I migrated from 1.5. |
43 |
|
44 |
Gentoo has many arches and the more popular a package is, the more |
45 |
bugreports will come, the harder it will be to mark it stable. Firefox |
46 |
is especially hard to maintain because users use it so very much. |
47 |
|
48 |
> Python releases are often behind, and not mentioning KDE 4, which is |
49 |
> even default on 8.10 Kubuntu and on Gentoo was still hardmasked last |
50 |
> time I checked (but probably Gentoo is just right in this respect, |
51 |
> everyone keeps telling me to wait before digging into KDE 4). |
52 |
|
53 |
Python is a special case. Portage (emerge and friends) use it, so we |
54 |
always try to have as few bugs as possible in the versions that are put |
55 |
into the tree. Kde 4.1 is broken, compared to 3.5.9/10. I tried it and I |
56 |
don't want it. The problem we have now is that 3.5.10 is starting to |
57 |
bitrot, so we'll probably *have* to mark 4.2 stable. |
58 |
|
59 |
> I fully understand that there are good reasons for that, and that the |
60 |
> meta-distribution status of Gentoo makes harder to check packages (and |
61 |
> also that the Ubuntu folks wildly release unstable stuff... firefox 3 |
62 |
> rc in 8.04, for example). I just feel that (stable) Gentoo is |
63 |
> actually a bit *behind* the average Linux distribution in its |
64 |
> revisions of software. |
65 |
|
66 |
You asked for stable, you got it. We're usually faster than Debian |
67 |
stable though. |
68 |
|
69 |
> Most importantly, I also feel that that's something new: when I first |
70 |
> installed my system, more than 4 years ago, I felt it was *ahead*. |
71 |
|
72 |
I did too, but then I was coming from Windows, so that's hardly |
73 |
surprising :-) |
74 |
|
75 |
No, seriously it didn't take long for me to go ~x86. I think it was |
76 |
ati-drivers (oh noez!) and keeping them in sync with xorg-server that |
77 |
drove me to it. |
78 |
|
79 |
> I wonder if it's due just to the sheer increase of work required to |
80 |
> test packages, or if there are decisions behind that (or if it's just |
81 |
> me having false memories). |
82 |
|
83 |
The amount of work has something to do with it, you (users) can help |
84 |
there by filing stable requests if you see a package that you feel has |
85 |
been ~arch for too long. We do react to nudges. Most of us, anyway. |
86 |
|
87 |
/PA |
88 |
|
89 |
[1] It wouldn't really be much fun being a dev for Gentoo if we didn't |
90 |
have the bestest users evers. Srsly :-). If you look at how many bug |
91 |
reports there are and how many are at least partially solved by users |
92 |
before a dev gets to it, it's quite humbling. Sometimes I can spend |
93 |
hours being a commit-monkey for users who've posted bugreports that |
94 |
makes solving the bug a matter of fifteen minutes, tops. |