1 |
It seems like there has been a lot of discussion here that indicates |
2 |
people are happy with the way it is. There seems to be differences in |
3 |
how packages are updated based on their purpose - desktop packages |
4 |
move very fast, a lot of server infrastructure moves more slowly. It |
5 |
seems like the "best" solution is already in place for the different |
6 |
usecases. |
7 |
|
8 |
If you hadn't noticed this, you may want to go look. I'm not sure if |
9 |
it's more due to choices made upstream or choices the maintainers |
10 |
make. |
11 |
|
12 |
|
13 |
On Sat, Jul 29, 2017 at 1:08 PM, Christopher Head <chead@×××××.ca> wrote: |
14 |
> I’m a stable user when I can be. I use Gentoo for the configurability, |
15 |
> not for instant access to the newest versions of things. |
16 |
> |
17 |
|
18 |
It seems to be understood in this discussion why some people use ~arch |
19 |
exclusively, but I would like to explain that it is a pattern I have |
20 |
seen very well myself; typically it's possible to solve bugs by |
21 |
keywording in the unstable package and being done with it. When |
22 |
problems related to unstable keywording arise it's usually because a |
23 |
system isn't completely unstable, which despite the name is very |
24 |
stable on Gentoo. |
25 |
|
26 |
It needs to be pointed out that all software in Portage is very new |
27 |
compared to other distributions. Stable and unstable packages work |
28 |
well together because they're closer in time to each other than other |
29 |
distributions respective categories. In fact, stable packages are *so |
30 |
new* compared to other distributions I think a more constructive |
31 |
question to ask is whether or not Gentoo should start retaining older |
32 |
packages for better interoperability with projects whose developers |
33 |
use non-Gentoo distributions.[1] |
34 |
|
35 |
On a distribution like Ubuntu or Debian a developer working for a |
36 |
software firm might wish to use, say, the very latest version of Ruby |
37 |
and Rails. To do this they might pull down the source release and then |
38 |
try to compile it themselves. This used to be the same as opening a |
39 |
portal to dependency hell,[2] but it's gotten better, and we assume |
40 |
the developer gets it installed to their home directory. They spin up |
41 |
a website with the new features and are done with it. The rest of |
42 |
their stack is whatever was in the package manager and might be, |
43 |
comparatively, very old. If they do this for more pieces of software - |
44 |
like if it is done on a developer's workstation, and not a single |
45 |
purpose server - eventually packages will start conflicting and things |
46 |
like containers and single purpose virtual machines start making |
47 |
sense. |
48 |
|
49 |
On Gentoo, the newest software is just there, and it's updated |
50 |
frequently enough that you never have to jump through breaking |
51 |
changes. Most people I have met that use Gentoo use it because they |
52 |
need lots of new software, or need to customize things in ways that |
53 |
are hard to do on other distributions. These people tend to realize |
54 |
that even if they run stable, those stable packages would probably be |
55 |
considered unstable on another distribution. |
56 |
|
57 |
R0b0t1. |
58 |
|
59 |
|
60 |
[1] Personally I don't think that would be a useful thing to do, I |
61 |
just install it in an Ubuntu or Debian VM if I want to play with that |
62 |
project. A lot of issues that exist in this regard are hardcoded paths |
63 |
and other things that come from the design of Ubuntu and Debian. |
64 |
|
65 |
[2] Ubuntu seems to keep their packages more up to date than they used |
66 |
to, because I remember having to compile 2-3 intermediate packages to |
67 |
get something to the newest version a couple of times. Debian still |
68 |
typically has very old software in their package repository. |