Bug#649514: nvidia-cg-toolkit: should this package be orphaned?
Mathieu Malaterre
mathieu.malaterre at gmail.com
Thu May 24 15:16:37 UTC 2012
On Thu, May 24, 2012 at 3:47 PM, Andreas Beckmann <debian at abeckmann.de> wrote:
> On 2012-05-24 14:57, Mathieu Malaterre wrote:
>> git clone git://github.com/anbe42/nvidia-cg-toolkit.git
>> cd nvidia-cg-toolkit
>> git checkout prepare-3.1v2
>> make -f ./debian/rules get-orig-source
>> mv nvidia-cg-toolkit_3.1.0013.orig-amd64.tar.gz
>> nvidia-cg-toolkit_3.1.0013.orig-i386.tar.gz ..
>> dpkg-buildpackage -rfakeroot -us -uc
>> make[1]: Entering directory `/tmp/nn/nvidia-cg-toolkit'
>> /usr/bin/make -C amd64/local/Cg/examples/Tools/cginfo/ \
>> CG_INC_PATH=/tmp/nn/nvidia-cg-toolkit/amd64/include
>> make: Entering an unknown directory
>> make: *** amd64/local/Cg/examples/Tools/cginfo/: No such file or
>> directory. Stop.
>
> That's not a proper unpacked source directory.
> But yes, we should probably add something to README.source explaining
> how to build from git.
>
> We could also add
>
> override_dh_autobuild:
> test -d i386 && test -d amd64 || \
> ( echo source not unpackaged ; exit 1 )
>
> I don't know a good replacement for the mergwithupstream mode from
> svn-buildpackage. How can something similar be done with git?
> I don't really want to add the unpacked blobs to git (at least as long
> as pristine-tar does not cope well with multiple source tarballs).
>
> As I only tested the build process with pbuilder (pdebuild) this was
> working fine for me because it first creates a source package (warning
> about a lot of ignored deletions) and thereafter has a complete source
> tree ...
Using pbuilder did work for me. I have no idea how to work around this anyway.
All I get is this:
$ lintian -I nvidia-cg-toolkit_3.1.0013-1_amd64.changes
I: libcggl: unused-override dev-pkg-without-shlib-symlink
I: libcg: unused-override dev-pkg-without-shlib-symlink
N: 11 tags overridden (2 warnings, 9 info)
I do not mind leaving those extra lintian overrides, but anyone knows
why there were here in the first place ?
>> trailing comma is not well handled by BTS, see extra '(u)' for instance at:
>> http://packages.qa.debian.org/k/khronos-opencl-headers.html
>
> I'd consider this a bug in the PTS. dch had similar problems but should
> have been fixed recently.
ok, I did not know that.
> PS: Miguel and Mathieu, are you subscribed to pkg-nvidia-devel@ ? So we
> could drop a few Cc:s
Yup, I just subscribed minutes ago.
Miguel any comments ?
--
Mathieu
More information about the pkg-nvidia-devel
mailing list