ktl/kernel_series: package.feeder is deprecated and no longer used
We have deprecated package.feeder in favour of package.depends but
retained an alias for live users. All of those are now updated so we
can now drop this support.
Signed-off-by: Andy Whitcroft <email address hidden>
Add a new set of relationships between packages in a source (see below
for a full set). These allow tooling to identify ordering, privacy, and
signing requirements for the various packages. We also add a new
package-relationships table with names relationship data. This allows
sources to indicate which semantics they have; for example for a source
converted to have a "linux-generate" package.
Each package has a set of associated keys as below:
adjunct: <true/false> -- the package is an adjunct package, a package
which must be uploaded to the private build PPA;
ancillary-for: <ptype> -- the package type of the package we are an
ancillary package for, the package in which this package is contained
and by which it is versioned;
depends: <ptype> -- the package type of the package we depend on, one
which much complete building before we can build;
signing-to: <ptype> -- indicates a package generates a custom signing
tarball and points to the package type which will ultimatly consume the
signed objects; and
signing-from: <ptype> -- indicates a package which consumes a custom
signing tarball and points to the package which generates it.
Signed-off-by: Andy Whitcroft <email address hidden>
ktl/kernel_series: package_lookup -- add lookup by type
We always intended this interface to be a by type lookup, but it ended
up using the key which switched to be the package name. Add a type=
parameter to select by type.
Signed-off-by: Andy Whitcroft <email address hidden>
ktl/kernel_series: clean up 'defaults' handling and elide assets
We are currently cheating in kernel-series using an assets key within
the defaults section to hold YAML fragments we are going to reuse. This
leads us to carry those defaults in all series data blocks even though
they are never used. Elide them formally from the data. Also take the
oppotunity to clean up the handling of defaults which we pass around
pointlessly. Use them directly from the KernelSeries object. Finally
prepare for the possibility that we may wish to have defaults at other
levels by using the 'series' key within defaults if it is present; thus
retaining backwards compatibility.
Signed-off-by: Andy Whitcroft <email address hidden>
swm-jira-sync: fix issue tag handling in the face of changes
When a Jira issue is replaced we will incorrectly add the new issue tag
without removing the old. Add support for detecting excess issue tags
and remove them. Add the appropriate new issue tag when needed.