Re: remove an installed rpm during rpm -i

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

 



On Thu, 4 Dec 2008, devzero2000 wrote:

> On Thu, Dec 4, 2008 at 10:30 AM, Panu Matilainen
> <pmatilai@xxxxxxxxxxxxxxx> wrote:
> > On Tue, 25 Nov 2008, devzero2000 wrote:
> >
> >> On Mon, Nov 24, 2008 at 8:36 PM, Seth Vidal
> <skvidal@xxxxxxxxxxxxxxxxx>
> >> wrote:
> >>
> >>
> >>       On Mon, 24 Nov 2008, Valery Reznic wrote:
> >>
> >>       > Did you try to use Obsolete tag ?
> >>       >
> >>
> >> a couple of things about the whole thread
> >>
> >> 1. obsolete does not work when you're rpm -i'ing pkgs. Only rpm -U
> >>
> >> 2. modifying the rpmdb from a %post or %pre is an unsafe idea and
> >> prone to
> >> causing things to "end badly".
> >>
> >>
> >> It depends. If Jeff tell it is possible to have an "RPM bundle" (e.g.
> an
> >> rpm that install rpm(s) itself) then perhaps
> >> it is not so unsafe in first place. I personally have written un rpm
> that
> >> use yum for pull update.
> >
> > It "works", except when it doesn't. If you can make sure the bundle is
> > only ever installed in a transaction of it's own, without --root, it'll
> > probably work just fine. But "rpm from within rpm" does break in some
> > cases, for example certain chroot install scenarios and more esoteric
> > cases such as glibc futex implementation changing in the same
> transaction.
> > And the breakage will always happen when you least expect it ;)
> 
> I must therefore imagine that the ability to " concurrent access" of rpm
> it is wrong.
> 
> i.e. rpm --showrc
> .......
> 
> Features supported by rpmlib:
>     rpmlib(VersionedDependencies) = 3.0.3-1
>         PreReq:, Provides:, and Obsoletes: dependencies support versions.
>     rpmlib(ConcurrentAccess) = 4.1-1
>         package scriptlets may access the rpm database while installing.
> 
> For me it is only little documented. But if it is thus harmful, it can
> sure itself  be removed if you think so.
> 
> But sometime it is a useful feature (for me almost) . For example there
> exists cases of installation of RPM products proprietary that do not
> mandate rpm dependencies but that they work only  if installed in a
> precise order. A rpm bundle it is a way for solving this problem.

Didn't say it was not useful, I only warned about the very real pitfalls. 
Rpm bundles are likely to work just fine for the "download an rpm-bundle 
off a website and install with rpm -Uvh". The trouble enters in more 
complex cases, such as glibc (or rpm itself) getting upgraded in the same 
transaction.

 	- Panu -
_______________________________________________
Rpm-list mailing list
Rpm-list@xxxxxxxxxxxxx
http://lists.rpm.org/mailman/listinfo/rpm-list


[Index of Archives]     [RPM Ecosystem]     [Linux Kernel]     [Red Hat Install]     [PAM]     [Red Hat Watch]     [Red Hat Development]     [Red Hat]     [Gimp]     [Yosemite News]     [IETF Discussion]

  Powered by Linux