[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Some myths regarding apt pinning

On Sun, Jan 26, 2003 at 05:33:42AM +0100, Thomas Hood wrote:
> I guess there is some misunderstanding here.  What I called
> A.B.'s "worst case scenario" is the scenario in which libc6
> gets upgraded once from unstable ... to a buggy version ...
> and is subsequently not upgraded from there again even when
> a security fix is released.  As he pointed out, security
> fixes usually appear in unstable the same time they
> appear in stable, so the worst-case scenario is avoided if
> pinning is set up such that the "unstable" packages continue
> to be upgraded from unstable.

Unfortunately, this cannot in general be done with pinning.
Installing an unstable package does not guarantee that it continues
to be upgraded from unstable.

> preferences.  Suppose I set the priorities of distributions
> as follows
>     stable 900
>     testing 800
>     unstable 700

If you install foo from unstable, and then that version makes it
into testing, and you upgrade before a new version is available in
unstable, you will stay with testing.

Actually, in your previous message, you seem to understand this--not
sure why you neglect it here.  Maybe you meant that it works if you
include only stable and unstable in your sources.list.  But I doubt
many people do that.

Anyway, this makes pinning much less useful than it might be.


Reply to: