Good morning,
the lack of reaction gives me the impression that not everyone on this list
fully grasps the implication of the new RPM behaviour.
I (actually a colleague of mine) have construed a simple RPM package
"fileconflictstest" which has an example file in it, in this case a kernel
module file, which is already present in the filesystem as part of an already
installed system, but in another version, i.e. the files are different.
BAD:
====
INSTALLING the "fileconflictstest" package overwrites the file without
warning.
WORSE:
======
REMOVING the "fileconflictstest" package leaves the second version of the file
not restoring the original file!
I.e. an "rpm -V" on the original RPM package gives an MD5 error, of course.
BRAINDEAD:
==========
the switch "--fileconflicts" is dysfunctional.
Here's the proof:
<example>
[EMAIL PROTECTED] /root # rpm -qf \
/lib/modules/2.6.16.13-4-smp/kernel/drivers/infiniband/core/ib_core.ko
kernel-smp-2.6.16.13-4
[EMAIL PROTECTED] /root # rpm -V kernel-smp-2.6.16.13-4
[EMAIL PROTECTED] /root # rpm -qlpv fileconflictstest-1.0-1.x86_64.rpm
-rw-r--r-- 1 root root 84232 Jun 21
16:30 /lib/modules/2.6.16.13-4-smp/kernel/drivers/infiniband/core/ib_core.ko
[EMAIL PROTECTED] /root # rpm -ihv --fileconflicts
fileconflictstest-1.0-1.x86_64.rpm
Preparing... ########################################### [100%]
1:fileconflictstest ########################################### [100%]
[EMAIL PROTECTED] /root # rpm -qf \
/lib/modules/2.6.16.13-4-smp/kernel/drivers/infiniband/core/ib_core.ko
kernel-smp-2.6.16.13-4
fileconflictstest-1.0-1
[EMAIL PROTECTED] /root # rpm -V kernel-smp-2.6.16.13-4
S.5....T
/lib/modules/2.6.16.13-4-smp/kernel/drivers/infiniband/core/ib_core.ko
[EMAIL PROTECTED] /root # rpm -e fileconflictstest
[EMAIL PROTECTED] /root # rpm -V kernel-smp-2.6.16.13-4
S.5....T
/lib/modules/2.6.16.13-4-smp/kernel/drivers/infiniband/core/ib_core.ko
</example>
Conclusion: Corrupt RPM database.
Proof enough?
Please be aware that I am not blaming SUSE & Co. for it, this is a RedHat &
Co. invention. It is just that SUSE (as far as I can see it) seems not to be
aware of this.
I am still lacking an adequate expression for this, but somehow a Beavis &
Butthead quote crosses my mind, I don't know why:
"This sucks more than anything that ever sucked before."
Regards
Oliver
Am Mittwoch, 21. Juni 2006 16:05 schrieb Oliver Tennert:
> Am Mittwoch, 21. Juni 2006 16:01 schrieb Michael Schroeder:
> > On Wed, Jun 21, 2006 at 03:49:20PM +0200, Oliver Tennert wrote:
> > > do I understand it right that since RPM 4.3 or so file conflicts when
> > > installing several RPMs are not shown anymore but ignored?
> >
> > No, there's no change in the file conflict handling. Why do you
> > think there is?
>
> Because I am reading the CHANGELOG in /usr/share/doc/packages/rpm/:
>
> [...]
> 4.2.2 -> 4.2.3:
> - bump rpm and popt versions to insure "newer".
> - change default behavior to resolve file conflicts as LIFO.
> - add --fileconflicts to recover rpm traditional behavior.
> [...]
>
> That means, two RPMs with say 1 file the same in both packages will install
> without warning, the last package overwriting the file of the first.
>
> Oliver
--
"The bland leadeth the bland and they both shall fall into the kitsch."
--
__
________________________________________creating IT solutions
Dr. Oliver Tennert
Senior Solutions Engineer
CAx Professional Services
science + computing ag
phone +49(0)7071 9457-598 Hagellocher Weg 71-75
fax +49(0)7071 9457-411 D-72070 Tuebingen, Germany
[EMAIL PROTECTED] www.science-computing.de
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
