| View previous topic :: View next topic |
| Author |
Message |
soehest n00b

Joined: 30 Aug 2007 Posts: 12
|
Posted: Mon Aug 06, 2012 8:17 pm Post subject: State of gentoo-sources |
|
|
Being a long time Gentoo user one thing has always puzzled me. The state of gentoo-sources. Having used some time tracking down load average trouble (a regression which was introduced in 3.2.17 and recently fixed in 3.5-rc7) In my search i did not find kernels marked as stable in gentoo-sources to me more stable than ones marked as testing. As it is now new users will install kernel 3.3.8 which is the newest kernel that has been marked stable. But Kernel 3.3.8 has been marked as EOL on http://kernel.org and has not received any updates in two months (last update 2012-06-01) Browsing at http://packages.gentoo.org/package/sys-kernel/gentoo-sources is really confusing. There will be several stable kernels in the same branch:
3.0.17-r2
3.0.35
3.2.1-r2
3.2.21
All are marked as stable and all have newer versions available on http://kernel.org
3.1.10-r1
Not even to be found on http://kernel.org
Having previously tried another source based distribution i noticed that they will use only a few kernels and will always use the lastest stable kernel from http://kernel.org as their "stable" kernel new users will install. I see the logic in using the newest (marked on http://kernel.org as stable) as the default stable kernel. Why not do the same for Gentoo? It would make it easier to maintain gentoo-sources by only supporting a few branches. And this would prevent users from installing EOL kernels
Best Regards |
|
| Back to top |
|
 |
John R. Graham Administrator


Joined: 08 Mar 2005 Posts: 6594 Location: Somewhere over Atlanta, Georgia
|
Posted: Mon Aug 06, 2012 8:28 pm Post subject: |
|
|
In Gentoo, "stable" has a specific meaning, namely that an arch tester has checked it and (typically) there are no confirmed/unresolved bugs outstanding against the exact category/package-version-revision for at least 30 days. Just because upstream says it's there new "stable" release doesn't meant it's stable in Gentoo. That said, resources are sometimes an issue and sometimes things don't move as quickly as we'd like.
- John _________________ This space intentionally left blank. |
|
| Back to top |
|
 |
Suicidal l33t


Joined: 30 Jul 2003 Posts: 933 Location: /dev/null
|
Posted: Tue Aug 07, 2012 2:45 am Post subject: |
|
|
I hear ya John, Ive been bitten by quite a few bugs that were considered stable by kernel.org. In those cases my only recourse was to revert to a stable kernel. _________________ Who controls the past now, controls the future
Who controls the present now, controls the past. |
|
| Back to top |
|
 |
mv Advocate


Joined: 20 Apr 2005 Posts: 3226
|
Posted: Tue Aug 07, 2012 5:50 am Post subject: |
|
|
| John R. Graham wrote: | | In Gentoo, "stable" has a specific meaning |
But somehow this meaning is broken for kernels: It means that in practice even stable users should be recommended to run ~ARCH kernels if they do not want to risk unreported security issues.
Actually, in other projects it is different: If possible and not too much out of the timeline, maintainers attempt to file stable requests for packages which are considered stable upstream (it is one of the policies of gentoo to follow upstream as far as it makes sense).
So IMHO, it would make sense trying to stabilize kernels with long-time support and to quick-stabilize their updates (since this involves usually [undocumented] security issues or essential fixes): This seems like the proper thing for stable users. Having kernels as stable which are unsupported by upstream simply makes no sense: These kernels have very likely undocumented by fixed by upstream security issues. |
|
| Back to top |
|
 |
mv Advocate


Joined: 20 Apr 2005 Posts: 3226
|
Posted: Tue Aug 07, 2012 5:53 am Post subject: |
|
|
| Suicidal wrote: | | Ive been bitten by quite a few bugs that were considered stable by kernel.org. In those cases my only recourse was to revert to a stable kernel. |
It could have happened also the opposite. It is good to have some alternatives, but stable should be those kernels where all known security issues are fixed. And only upstream knows which kernels these are. |
|
| Back to top |
|
 |
|