1 |
Thanks for all the feedback, everyone. |
2 |
|
3 |
> Markus Hauschild: |
4 |
> If you really want to test ~arch packets you don't necessarily need |
5 |
> ~arch stages to download, you can just switch your Installation to |
6 |
> ~arch and then file bugs etc. |
7 |
|
8 |
That's what we did, and what generated the ~tarball suggestion. |
9 |
|
10 |
> Alex Howells: |
11 |
> Look at it this way: by running ~arch whilst *not* a Developer or |
12 |
> Arch Tester you're having a very limited impact, or possibly a |
13 |
> negative one. Getting onto the 'track' of contributing to the project |
14 |
|
15 |
Contributing...I just tried a couple of suggestions? They seem good to |
16 |
me. |
17 |
|
18 |
It isn't preference or 133t-ness. There are technical issues with the |
19 |
user machines and desktop lust. |
20 |
|
21 |
I'm not saying "change your ways" but rather "tarball ~stuff" to help |
22 |
sysadmins make their own design choices. Any choice is a balancing act |
23 |
of competing requirements. |
24 |
|
25 |
> They've got no clue what it means, then they bitch/whine |
26 |
> when they hit ABI issues or other problems and blame Gentoo. |
27 |
|
28 |
Not in this discussion? All I want is a cleaner way to install ~arch. |
29 |
Put all the warning stickers you want. I agree it is *not* for average |
30 |
users. |
31 |
|
32 |
Many feel Debian unstable is the more stable branch, because it swallows |
33 |
upstream bugfixes. Debatable; can depend on the system spec. Debian |
34 |
focuses too much on servers -- it ought to fork a desktop branch, if you |
35 |
ask me. Some Debian distros have done just that. Anyway the point is, |
36 |
there can be legit reasons to run unstable; reasonable people can |
37 |
differ. |
38 |
|
39 |
There is lag between upstream package releases and distro adoption. |
40 |
Typical scene: an upstream package advertises "now more stable!" but |
41 |
the distro takes a year or two rolling it in. Worse scene: upstream |
42 |
package advertises "now supports your hardware!" but again, the distro |
43 |
takes 1-2 years. |
44 |
|
45 |
So the dilemma: which branch is really the more "stable"? The one that |
46 |
the distro calls stable, or the one with all the latest from upstream? |
47 |
There is no one answer of course. Obviously a release engineering |
48 |
statement on the matter is going to be different from another viewpoint. |
49 |
|
50 |
I follow release engineering's worries about user install procedures, |
51 |
and that's legit. But I am a sysadmin, unafraid of reasonable breakage |
52 |
that I can fix. I would not recommend average people install ~arch any |
53 |
more than you would. All I'm saying is ~tarballs would be nice for |
54 |
experts. |
55 |
|
56 |
My job reviews aren't based on making Gentoo penetrate this or that |
57 |
market sector but making computers work. I don't have the luxury of |
58 |
explaining to folks that "the distro will take care of it in 1-2 years" |
59 |
or endlessly fiddling with custom package selections ("apt-pinning" in |
60 |
Debian). Users want me out of their cubes, fast. |
61 |
|
62 |
> run ~arch with XFS on a desktop system that doesn't have a UPS |
63 |
|
64 |
Guilty as charged. Running Debian unstable on XFS for years, through |
65 |
dozens of storm blackouts, and zero data loss. Ext3 lost plenty of data |
66 |
before we gave up on it. Have no intention of using ext4, either. |
67 |
|
68 |
We should have UPSes, if only the bean counters would stop retorting |
69 |
that we've never lost data, so why do we need 'em...ha. |
70 |
|
71 |
(Good fstab tips: barrier, noatime, nodiratime...and /tmp and /var/log |
72 |
in tmpfs...) |
73 |
|
74 |
The consensus here is that we'll wait for beta release and install that |
75 |
with ~arch keyword. Lookin' forward to it. |
76 |
-- |
77 |
|
78 |
davecode@××××××××××.net |
79 |
|
80 |
-- |
81 |
http://www.fastmail.fm - One of many happy users: |
82 |
http://www.fastmail.fm/docs/quotes.html |
83 |
|
84 |
-- |
85 |
gentoo-releng@l.g.o mailing list |