← Back to team overview

linaro-release team mailing list archive

[Bug 724637] Re: FFe: dpkg 1.16.x: multiarch support

 

It's still early after FF, so if we can get this in for alpha-3,
approved. This should give us enough leeway for sorting out regressions.

Also, the earlier we get rid of this ia32-libs hack, the better. It has
been a constant source of obsolescence, bugs (missing dependencies), and
unfixed security issues.

** Changed in: dpkg (Ubuntu)
       Status: New => Confirmed

-- 
You received this bug notification because you are a member of Linaro
Release Team, which is a direct subscriber.
https://bugs.launchpad.net/bugs/724637

Title:
  FFe: dpkg 1.16.x: multiarch support

Status in “dpkg” package in Ubuntu:
  Confirmed

Bug description:
  Binary package hint: dpkg

  I crave the indulgence of the Ubuntu release team in permitting a
  post-feature-freeze upload of dpkg 1.16.x.  This is needed in order to
  have package manager support for multiarch packages in natty; it has
  not been uploaded before feature freeze specifically because the on-
  disk database format has not been finalized yet, and it's better to
  have the final on-disk format a little late than to have to transition
  everyone's data later.

  As a result, however, this will include a number of unrelated changes
  that are also staged in the dpkg git tree for upload.  I think that's
  preferred over cherry-picking multiarch here because holistic testing
  of the multiarch support is being done against the git trunk.

  Justification: this is a prerequisite for co-installability of
  x86+x86_64 (or armel+armhf, or...) packages on Ubuntu systems, a key
  component of https://blueprints.launchpad.net/ubuntu/+spec
  /packageselection-foundations-n-multiarch-support needed to let us get
  rid of the ia32-libs package from Ubuntu.

  Potential impact:  the new dpkg may eat everyone's systems.  But we'll
  probably find out about that pretty fast if it's going to happen, so
  no risk to the release itself as far as I can discern.

    [ Guillem Jover ]
    * Fix realloc usage on compat scandir() implementation.
    * Use DPKG_MAINTSCRIPT_PACKAGE environment variable as package name on
      dpkg-divert when no --package or --local options have been specified.
    * Do not allow versions starting with non-digit when doing strict parsing,
      warn otherwise.
    * Update dpkg(1) to note that --status-fd output does not contain newlines
      in error messages anymore (this was fixed in 1.15.0).
    * Add a new --status-logger option to dpkg, similar to --status-fd but
      instead invoke the command ourselves and feed the status information
      to its standard input. Suggested by Raphaël Hertzog.
    * Add missing space in update-alternative --set-selections output.
    * Add missing options to update-alternative --help output.
    * Count “conffile name is duplicated” for dpkg-deb warning count summary.
    * Improve and clarify strings for translation. Closes: #604914
    * Prefix all fatal error messages with “error: ”.
    * Do not check presence of update-rc.d in the PATH in dpkg, as it's not
      a program needed for dpkg correct operation.
    * Fix dpkg -GEO options on multiple versions of the same packages.
      Closes: #31141
    * Propagate --admindir to programs run from maintainer scritpts.
      Closes: #97076
    * Do not fail when trying to remove the root directory. This will only
      happen either on distributions where dpkg is a foreign package manager,
      or on artificial dpkg databases.
    * Always warn when parsing any package control data which does not have
      an Architecture field except for status and status log files when
      packages are not-installed or half-installed.
    * By default reject installing packages w/o an Architecture field. They
      now need --force-architecture, dpkg will still warn about them though.
    * Fix build failure when passing --disable-nls to configure.
    * Do not segfault on “dpkg -i --no-act”.
    * Add missing semicolon to the vsnprintf() compat declaration.
      Thanks to Robert Millan. Closes: #612203
    * On install for Ubuntu adjust the i386 GNU cpu name in cputable.
      Thanks to Colin Watson <cjwatson@xxxxxxxxxx>. Closes: #611741

    [ Raphaël Hertzog ]
    * Fail properly when debian/source/format is empty. Closes: #600854
    * Add new deb-src-control(5) manual page documenting the debian/control
      file contained in source packages.
      - it documents the X[SBC]- prefix. Closes: #476335
      - it documents the VCS-* fields too. Closes: #483119
      Thanks to Oxan van Leeuwen <oxan@xxxxxxxxxxxxxxxxx> who wrote it
      as part of the Google Code In program.
    * Enhance dpkg-shlibdeps to not fail immediatly when a library is not found.
      Instead continue and fail after all problems have been reported. Thanks
      to Chris Baines <cbaines8@xxxxxxxxx> for the patch. Closes: #596841
    * Fix dpkg-source to not list Debian packaging files as modified
      upstream files in Format "1.0" when unpacking to a non-standard
      directory.

    [ Jonathan Nieder ]
    * Remove support for use of synchronous sync(2), due to its pernicious
      side-effects and to ease maintenance.

    [ Updated programs translations ]
    * German (Sven Joachim).
    * Portuguese (Miguel Figueiredo).
    * Spanish (Javier Fernandez-Sanguino).
    * Swedish (Peter Krefting).

    [ Updated man page translations ]
    * German (Helge Kreutzmann).
    * Swedish (Peter Krefting).

    [ Updated scripts translations ]
    * German (Helge Kreutzmann).
    * Swedish (Peter Krefting).

    [ Updated dselect translations ]
    * Spanish (Javier Fernandez-Sanguino).