1 |
Nuitari posted <Pine.LNX.4.63.0510141023280.8162@××××××××××××××××.net>, |
2 |
excerpted below, on Fri, 14 Oct 2005 10:24:11 -0400: |
3 |
|
4 |
>> Are u sure you want your whole system built out of the testing (~amd64) |
5 |
>> "release" of gentoo? |
6 |
> |
7 |
> My laptop runs with ~amd64 and it's the amd64 system I had the least |
8 |
> problems with... |
9 |
|
10 |
Likewise here, with my dual Opteron workstation. |
11 |
|
12 |
While I'm not going to say any particular Gentoo user should be running |
13 |
stable or ~arch (testing), amd64 is a bit different than x86 regarding |
14 |
stable. Altho the degree to which this is still true is fading, it's |
15 |
still true to some extent that the folks likely to be running amd64 in the |
16 |
first place are more likely to be leading-edge type folks. Thus, it's |
17 |
likely that a higher proportion of amd64 users use ~arch than x86 users, |
18 |
which corresponds to a more reliable ~amd64 than ~x86, because it has been |
19 |
better tested as there are more folks running it. (One way to ensure this |
20 |
even further would be to delay upgrades on critical packages by 3-4 days |
21 |
after they first appear in the upgrade list, watching to see if there's a |
22 |
bunch of problem reports on them. Such packages would certainly include |
23 |
gcc/glibc/binutils/portage/baselayout/xorg, and your X environment of |
24 |
choice, kde/gnome/whatever.) |
25 |
|
26 |
Correspondingly, stable amd64 users /may/ at times have /more/ troubles |
27 |
than ~amd64, because stable will by definition mean older versions, and |
28 |
with fewer folks running it, there might be occasional compatibility |
29 |
problems between older packages and something freshly marked stable -- |
30 |
which passed the tests to stable because everyone running it was running |
31 |
newer ~arch versions of whatever it had problems with. |
32 |
|
33 |
That said, ~arch /will/ at times have issues, particularly if you upgrade |
34 |
as soon as something gets marked ~amd64. As I mentioned, one way to avoid |
35 |
this would be waiting a couple days after a portage says a package is |
36 |
available, to upgrade to it, for packages which are either generally |
37 |
critical, or critical to you. |
38 |
|
39 |
Here, as mentioned, I run a ~amd64 system. I even unmask and merge |
40 |
masked-for-testing packages on occasion. Right now, for instance, I'm |
41 |
running gcc-4.0.1 as my main compiler, using eselect compiler to switch |
42 |
back to gcc-3.4.4 for stuff (like xorg) that doesn't yet compile with |
43 |
gcc-4.x. As part of that process, I also had to unmask newer versions of |
44 |
binutils with fixes for gcc4, and I'm running a still masked glibc with |
45 |
gcc4 fixes as well. I haven't yet tried the kde-3.5 alpha packages, but I |
46 |
did run the kde-3.4 betas, which as betas were never unmasked, and will |
47 |
probably do the same with kde-3.5 at some point. I tried the still masked |
48 |
modular-xorg stuff, but couldn't get it to run, so dropped back to the |
49 |
last monolithic xorg-x11 snapshot build, now xorg-x11-6.8.99.15-r4, as of |
50 |
last portage tree update a few days ago (I'll do another tonight, and may |
51 |
try kde-3.5 after that). I often run portage and baselayout snapshots |
52 |
before they are keyworded ~amd64 as well, altho the versions I'm running |
53 |
now (portage-2.0.53_rc5 and baselayout-1.12.0-pre9-r1), while rc/pre and |
54 |
therefore unlikely to ever go stable, are ~amd64 keyworded and not masked. |
55 |
|
56 |
As I said, far be it from me to tell someone they should run |
57 |
unstable/testing/~arch, but for those willing to work around the |
58 |
occasional issue, ~amd64 is probably safer than ~x86 would be, and for |
59 |
specific packages at specific times, has proven safer even than stable |
60 |
amd64. Again, for those wanting to try it but still a bit cautious, |
61 |
consider running ~amd64, but waiting a couple days after critical packages |
62 |
show up on the upgrade list, before trying them, to see if anybody "stupid |
63 |
enough to try them first" <g> runs into issues. (I can say that since I'm |
64 |
"that stupid"! <g>) |
65 |
|
66 |
-- |
67 |
Duncan - List replies preferred. No HTML msgs. |
68 |
"Every nonfree program has a lord, a master -- |
69 |
and if you use the program, he is your master." Richard Stallman in |
70 |
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html |
71 |
|
72 |
|
73 |
-- |
74 |
gentoo-amd64@g.o mailing list |