Control: fixed -1 6.4.0-2
Control: tags -1 pending

Hi Andreas,

Thank you for bringing this issue to my attention. A fixed package has
been uploaded to mentors and is waiting for sponsorship. (#723991)

On 09/21/2013 08:48 PM, Andreas Beckmann wrote:
> If the transtional packages are not installable, the package won't
> migrate to testing.
> The Breaks/Replaces against -6.2.1 need to be versioned as well.
> But do you really need transitional packages for a shared library?

As far as I know transitional packages are standard procedure when
packages are renamed. Taking your comments about osgearth (#723980) in
consideration, it seems transitional packages are not needed for shared

The API will be incompatible with the one of the old package, so it's
wrong to still provide the old package. The old API is not available in
the new package.

> Does this even work correctly?

With the transitional packages local builds of MapCache with MapServer
support will not be removed because its (unversioned) dependency on
libmapserver will still be satisfied. While the actual libmapserver is
not available anymore.

Leaving out the transitional packages, MapCache would remain functional
if only libmapserver is installed. It will not be considered for update
if the new libmapserver1 is not pulled in via any of the other mapserver

The case for MapCache is not an issue for Debian as it hasn't been
uploaded yet. But the packages are already part of the UbuntuGIS PPA.

> Wouldn't a transition with binNMUs on all rdeps be better?

If there were rdeps in Debian it would. But there fortunately aren't any
yet, but soon MapCache will become the first reverse dependency in Debian.

> If the library gained a stable API/ABI, you can probably close #327853.

Unfortunately MapServer still has no stable API, this is planned for the
7.0 release though.

> Same problem for the transitional ruby packages:

Should be fixed in the package on mentors.

> Cheers,
> Andreas

Kind Regards,


GnuPG: 0xE88D4AF1 (new) / 0x77A975AD (old)

Pkg-grass-devel mailing list

Reply via email to