[Acked] [kteam-tools][PATCH] git-build-kernel: do not expect orig tarball for debian "native" packages

Andy Whitcroft apw at canonical.com
Fri Aug 10 15:59:36 UTC 2018


On Tue, Aug 07, 2018 at 12:22:23PM -0700, Kamal Mostafa wrote:
> Determine whether or not to expect an orig tarball based on the file
> debian/source/format, as follows:
> 
>   - if the file does not exist:  NO orig tarball
>   - if the file contains "native" e.g. "3.0 (native)":  NO orig tarball
>   - otherwise:  YES orig tarball
> 
> This eliminates the need for name-based special cases for -meta and -signed
> and allows for the construction of main kernel source packages when an
> orig tarball has not (yet) been published.
> 
> Signed-off-by: Kamal Mostafa <kamal at canonical.com>
> ---
>  git-build-kernel/git-build-kernel | 9 ++++-----
>  1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/git-build-kernel/git-build-kernel b/git-build-kernel/git-build-kernel
> index ad6c3c5..831c0af 100755
> --- a/git-build-kernel/git-build-kernel
> +++ b/git-build-kernel/git-build-kernel
> @@ -142,11 +142,10 @@ touch "$WORKDIR/building"
>  trap "rm -rf $WORKDIR/building $BUILDDIR" 0
>  
>  [ $do_source_pkg = 1 ] && {
> -    ### Special case for linux-signed and linux-meta packages, which
> -    ### have no .orig tarball
> -    [ ${SRCPKG/-signed} != $SRCPKG ] && no_orig_tarball=1
> -    [ ${SRCPKG/-meta} != $SRCPKG ] && no_orig_tarball=1
> -    if [ "$no_orig_tarball" = 1 ]
> +    deb_format="$(git show $GITBRANCH:debian/source/format 2>/dev/null)"
> +    [ -z "$deb_format" ] && no_orig_tarball=1
> +    [ "${deb_format/native/}" != "$deb_format" ] && no_orig_tarball=1
> +    if [ "$no_orig_tarball" == 1 ]
>      then
>  	srcpkg_debopts="--no-tgz-check $srcpkg_debopts"
>      else

Much nicer.  No idea if the huristic is completely correct but
this will cover the the cases which I expect to hit in the field.
Well other than the time when don't have an orig tarball in devel
for linux.  There I think we will have to train those guys to
flip the version to 3.0 (native) for that period.  Regardless that
problem exists before this change:

Acked-by: Andy Whitcroft <apw at canonical.com>

-apw




More information about the kernel-team mailing list