1 |
On Sun, Dec 4, 2011 at 8:44 AM, Harry Putnam <reader@×××××××.com> wrote: |
2 |
> One point no one has mentioned and I've wondered from time to time |
3 |
> myself is whether one can expect gentoo to continue into the future |
4 |
> for a long while, as compared to the likely hood of opensuse or maybe |
5 |
> debian that has been around a very long time. |
6 |
> |
7 |
> It seemed at one time a year or so ago that gentoo's longevity was |
8 |
> questionable. (Possibly my own mis-perception) |
9 |
|
10 |
From my rudimentary gatherings while reading related blogs and |
11 |
historical perspectives, there are just two or three people who have |
12 |
been at the core of Gentoo for a very, very long time. Gentoo has long |
13 |
been the work of many hands, but these guys have been around the block |
14 |
a far more times. I don't know how well Gentoo would fare if one or |
15 |
three of them were to drop off the face of the earth. |
16 |
|
17 |
> For the OP, a few posters have mentioned that under gentoo, every |
18 |
> thing is compiled from scratch, but it was not made clear that it |
19 |
> happens again and again at most updates. |
20 |
> |
21 |
> No one has made clear that there is a very HUGE amount of time sunk |
22 |
> into compiling absolutely everything. |
23 |
|
24 |
I'd say "bull", but that depends *greatly* on your hardware. When I |
25 |
talk about Gentoo with my friends, they admit to having tried it, but |
26 |
then say it took them a long, long time to build a system on their |
27 |
486. You don't want to run Gentoo compiles on a 486. You probably |
28 |
ought not to run Gentoo compiles on any x86 processor older than an |
29 |
Athlon64 or Intel Core chip. |
30 |
|
31 |
For me, an emerge -e @world takes somewhere between four and ten |
32 |
hours, depending if it's the eight-core Xeon box or the quad-core |
33 |
Phenom box. As others noted in the build-speed optimization thread, |
34 |
it's pretty trivial to tune the system so that it doesn't impact many |
35 |
(most?) normal user activities, and can go on in the background. |
36 |
Otherwise, a full system rebuild isn't much more time consuming than |
37 |
something like a dist-upgrade on Debian or Ubuntu. |
38 |
|
39 |
There are factors you can tweak to go one way or the other, too. You |
40 |
might use bindists for chromium, firefox, thunderbird, xulrunner, |
41 |
libreoffice... That'd probably cut my Phenom system's compile time by |
42 |
about a quarter. I know installing a full KDE package set would |
43 |
*increase* build time on my system by about the same. |
44 |
|
45 |
The vast majority of the time, you're not building a full package set, |
46 |
but just ten or eleven packages. (if you let things slip a week or |
47 |
two, like I'm apt to do) |
48 |
|
49 |
> |
50 |
> A single update, if one lets updating slip a bit, can literally take |
51 |
> days to compile. |
52 |
|
53 |
AFAIK, it can't take longer than an emerge -e @world, which I described above. |
54 |
|
55 |
> And more days to reconfigure so that everything works |
56 |
> again. |
57 |
|
58 |
That seems very unusual, unless by "a bit" you're talking on the order |
59 |
of six months. |
60 |
|
61 |
|
62 |
-- |
63 |
:wq |