Re: Careful with dist-upgrade in unstable at the moment
Diederik de Haas - 22.12.22, 13:26:47 CET:
> On Thursday, 22 December 2022 09:16:50 CET Martin Steigerwald wrote:
> > > I find it way easier to have apt reduce the problem riskless
> > > first.
> > > It's a shorter list of actions to review.
> >
> > Good argument. My argument is that in the usual situation trying
> > "apt
> > full-upgrade" first will save me one command. With "apt upgrade" I
> > often enough would have to use "apt full-upgrade" afterwards.
>
> IMO that indicates that the 'state' of your packages could be
> improved. I *rarely* have to do a full-upgrade to get things fully
> upgraded. And when not all packages get upgraded, that usually means
> something 'special' is going on, like now with the Qt transition.
Well, I still thought about the "apt-get upgrade" scenario even though I
use "apt upgrade" meanwhile. Often enough there are new packages to
install. I did not really test how often I would have to use "apt full-
upgrade" instead of "apt upgrade". It might be considerably less often
due to apt installing new packages automatically.
> On Wednesday, 21 December 2022 11:42:17 CET Diederik de Haas wrote:
> > I think they above quoted script is absolutely horrific.
>
> I made that statement for 2 reasons:
> 1) It tries do a dist-upgrade 'at all cost' (imo ofc)
> 2) `dpkg --set-selection` completely messes up APT's 'database' wrt
> manually and automatically installed packages ... which in turn
> causes the need to full-/dist-upgrade.
I made no statement at all about this script. So in case you were
assuming that I somehow intended to justify it, I was not. Actually I
did not even carefully read through it to see what it does.
My statement was just about "upgrade" versus "full-upgrade".
Best,
--
Martin
Reply to: