Re: aptitude --mind-your-own-business option?
On Sun, Dec 03, 2006 at 01:05:29AM +0100, Florian Kulzer wrote:
> On Sat, Dec 02, 2006 at 11:27:46 -0900, Ken Irving wrote:
> > If it isn't "safe" to use apt-get once aptitide is installed, i.e.,
> > apt-get conflicts with aptititude, then couldn't this be handled in the
> > package setup as policy?
>
> It is perfectly safe to mix aptitude and apt-get. I will now perform a
> daring experiment to demonstrate this: I have here an up-to-date Debian Sid
> box, which I have upgraded daily with aptitude for the last year (before
> that I used apt-get, also daily). Now, what unspeakable horrors lie in
> wait for me if I install something with apt-get and run aptitude
> afterwards? There is only one way to find out...
>
> First of all, an md5sum of aptitude's list of installed packages (which
> includes the "auto" flag) - I want to be able to show that I can restore
> this state:
>
> $ aptitude search '~i' | md5sum
> 06f9da945e91f95eb2913ed081809159 -
> ...
> # apt-get install aiksaurus
> ...
> # aptitude dist-upgrade
> ...
> # apt-get dist-upgrade
> ...
> # aptitude --purge-unused purge aiksaurus
> ...
> $ aptitude search '~i' | md5sum
> 06f9da945e91f95eb2913ed081809159 -
>
> I rest my case.
I'll take your word for it. So once aptitude gets "synched" to my system,
I can use either aptitude or apt-get with impunity? That sounds reasonable.
But its initial behavior still strikes me as unreasonable. Would aptitude
be left in a broken state if it warned me of inconsistencies but still did
what I asked (e.g., install something in spite of unused packages)? Could
not this work be postponed to some time when it is, in fact, critical?
Ken
--
Ken Irving, fnkci@uaf.edu
Reply to: