Control: forcemerge 813455 -1

On Sun, 2016-07-03 at 02:42:55 +0100, Matthew W.S. Bell wrote:
> Package: dpkg
> Version: 1.18.7
> Severity: serious
> Justification: Prevents the installation of a packages/leaves package
> in a broken state.

> Installing a new version of the package imagemagick is failing:
> 
> #dpkg -i /var/cache/apt/archives/imagemagick_8%3a6.9.2.10+dfsg-2_all.deb 
> (Reading database ... 431189 files and directories currently installed.)
> Preparing to unpack .../imagemagick_8%3a6.9.2.10+dfsg-2_all.deb ...
[…]
> dpkg-maintscript-helper: error: directory '/usr/share/doc/imagemagick' 
> contains files not owned by package imagemagick:all, cannot switch to symlink
> dpkg: error processing archive 
> /var/cache/apt/archives/imagemagick_8%3a6.9.2.10+dfsg-2_all.deb (--install):
>  subprocess new pre-installation script returned error exit status 1
> Errors were encountered while processing:
>  /var/cache/apt/archives/imagemagick_8%3a6.9.2.10+dfsg-2_all.deb

> The current status of the package imagemagick from dpkg -l is:
> ii  imagemagick          8:6.9.1.2-1     amd64           image manipulation 
> programs -- binaries
> 
> The problematic output appears to be coming from 
> dpkg-maintscript-helper, invoked with the following dpkg environment
> variables, and command line:
> DPKG_ROOT=
> DPKG_MAINTSCRIPT_DEBUG=1
> DPKG_MAINTSCRIPT_ARCH=all
> DPKG_RUNNING
> _VERSION=1.18.7
> DPKG_MAINTSCRIPT_NAME=preinst
> DPKG_MAINTSCRIPT_PACKAGE=im
> agemagick
> DPKG_ADMINDIR=/var/lib/dpkg
> #dpkg-maintscript-helper dir_to_symlink /usr/share/doc/imagemagick 
> /usr/share/doc/imagemagick-6-common 8:6.9.2.10+dfsg-2~ -- upgrade 8:6.9.1.2-1 
> 8:6.9.2.10+dfsg-2
> 
> The problem seems to be that dpkg-maintscript-helper is expecting that
> the already installed version is of the same architecture as given in
> DPKG_MAINTSCRIPT_ARCH. It is unclear to me if this is a deficiency of
> dpkg-maintscript-helper or the caller.

I guess it's a bit of both. I consider the current behavior in
dpkg-maintscript-helper to be a bad default optimized for the wrong
thing, which I'll try to fix before checking that this will not incur
much fallout. But as it stands this is bad usage from the package
affected. Merging with the rest.

Thanks,
Guillem

Reply via email to