1 |
On Sat 20 August 2011 22:13:07 Pandu Poluan did opine thusly: |
2 |
> I hope someone can shed me some light here. |
3 |
> |
4 |
> I keep finding myself doing time-consuming emerges for my Gentoo |
5 |
> (virtual) systems (e.g., gcc-4.5.3, glibc-2.13, emerge -e, and so |
6 |
> on). So, I found myself wanting to build a so-called 'stage3.1' |
7 |
> tarball (i.e., a stage3 tarball *plus* the things I did all this |
8 |
> time). |
9 |
> |
10 |
> Now, my systems have different USE flags, depending on its usage. So |
11 |
> my question is: |
12 |
> |
13 |
> Can I just disregard the differences in USE flags for my stage3.1 |
14 |
> (e.g., just use the most-minimal amount of USE flags) and do an |
15 |
> emerge -avuND @system @world for every system having a different |
16 |
> set of USE flag? Or should I make one stage3.1 tarball for each USE |
17 |
> flag combination? |
18 |
|
19 |
Either way works. All you have here is a classic case of finding the |
20 |
sweet spot that is maximum commonality and minimum hassle to tweak it. |
21 |
|
22 |
Only you can define where that sweet spot is, as the answer relies on |
23 |
things like how much resources you have to re-compile, the number of |
24 |
re-emerging to be done, and how little (or much) tolerance you have. |
25 |
|
26 |
To get a real answer you'd have to give full details on your new |
27 |
tarball, USE flags, and how the actual machines using them differ. |
28 |
Then describe the impact of those changes and which bits you are happy |
29 |
with. I then doubt many people would bother reading and responding :-) |
30 |
|
31 |
Personally, I consider anything that needs glibc, gcc and the bulk of |
32 |
@system to be rebuild to be a PITA and I'd be making different |
33 |
tarballs for those once. But if the list of remerges is say 30 perl |
34 |
packages then I wouldn't bother and just stick with one tarball as |
35 |
that update is about 4 minutes worth of time. But that's just me. |
36 |
|
37 |
|
38 |
-- |
39 |
alan dot mckinnon at gmail dot com |