| 1 |
Hi, |
| 2 |
|
| 3 |
Michael Ulm wrote: |
| 4 |
> Our family computer has many different tasks to perform |
| 5 |
> and consequently has many packages emerged. Several of |
| 6 |
> these packages were not available for amd64 stable, so I |
| 7 |
> got the ~amd64 versions. This of course led to the need |
| 8 |
> to also pull in ~amd64 dependencies. And those dependencies |
| 9 |
> grow. |
| 10 |
|
| 11 |
What packages exactly are you talking about? I'm running a pretty stable system |
| 12 |
here and don't have a big package.keywords list (well, beside GNOME 2.12, |
| 13 |
modular X and the e17 cvs builds ;)). |
| 14 |
|
| 15 |
Just as hint: Use version-specific atoms. This will ensure that you don't always |
| 16 |
get the unstable version but only the version you really want. By keeping the |
| 17 |
version low, it will also significantly reduce the dependencies you have to |
| 18 |
keyword, in most cases. |
| 19 |
|
| 20 |
> 1) spending several hours checking out the ~amd64 packages I |
| 21 |
> emerged, to see if they are available in stable, and which |
| 22 |
> dependencies can then be brought back to amd64. As spare |
| 23 |
> time is a rather scarce resource for me, I don't want to do |
| 24 |
> this unless absolutely necessary |
| 25 |
|
| 26 |
This would of course be the cleanest solution. Move your package.keywords file |
| 27 |
somewhere else and run emerge -uDp world, then check if it would downgrade to a |
| 28 |
version which didn't work for you. I guess that most of these 50 entries you |
| 29 |
have wouldn't actually cause a downgrade to happen so it should reduce the |
| 30 |
effort pretty much. |
| 31 |
|
| 32 |
> 2) Go totally ~amd64. I am slightly worried about system |
| 33 |
> stability in this scenario. Every time the system hiccups |
| 34 |
> my wife tells me that this never happened in Windows... |
| 35 |
|
| 36 |
Although many people run ~amd64 without having any issues, I wouldn't suggest |
| 37 |
this if you don't like to fix problems and/or spend much time on administrating. |
| 38 |
|
| 39 |
> 3) Do nothing and hope for the best. |
| 40 |
|
| 41 |
Never change a winning horse. You could just stop doing emerge -uD world at all |
| 42 |
and only update packages if you know higher versions have new features you need |
| 43 |
or fixes a bug you hit. Instead of updating all packages, only run `glsa-check |
| 44 |
-f new` regularly to make sure your system doesn't get vulnerable. This is the |
| 45 |
way I do it on all my critical systems, and it works. The time I need to |
| 46 |
administrating is really minimal, about 10 min/mt. for a server and a |
| 47 |
workstation. I'd say this is even less than with windows ;) |
| 48 |
|
| 49 |
> 4) Go Kubuntu. I really like Gentoo, but if system administration |
| 50 |
> will take too much time, this may be my only path left :-( |
| 51 |
|
| 52 |
Disclaimer: I only made experiences with Ubuntu, so it might be that the |
| 53 |
following is not entirely true. |
| 54 |
|
| 55 |
I made the experience that if you want to use anything that is outside a small |
| 56 |
range of packages (the most popular tool for a task), you have to add various |
| 57 |
sources to your apt-get list to get those not-so-common packages. For some |
| 58 |
packages that means that you even have to go back to the debian sources, which |
| 59 |
aren't made for your Ubuntu system anyway. So you'll have to add things like |
| 60 |
Universe and so on to your source list to get the things done, which is pretty |
| 61 |
similiar to using a fully ~amd64 system. I'd personally rather go with option 2) |
| 62 |
than with 4), but it's up to you ;P |
| 63 |
|
| 64 |
Regards, |
| 65 |
|
| 66 |
-- |
| 67 |
Simon Stelling |
| 68 |
Gentoo/AMD64 Operational Co-Lead |
| 69 |
blubb@g.o |
| 70 |
-- |
| 71 |
gentoo-amd64@g.o mailing list |
Archives