585deac61c
The various post-release composes - two-week Atomic, Cloud, and Docker - all send out identical fedmsgs. The other commits in this PR at least ensure the fedmsgs sent after the compose is done will have a `compose_id` key, allowing disambiguation that way, but fedmsgs sent *before* the compose is done cannot be told apart. This should solve that, without inventing new message topics: we add a `short` key to the message content which specifies the 'shortname' of the compose (e.g. Fedora-Atomic), which is how we distinguish these composes from each other. We also remove a bunch of unused variables from these files, and revise the release-related variables as explained for the master branch in #485: `$DIST`, `$BRANCH` and `$BRANCHED` are all replaced by `$RELEASE` and `$RELEASE_TITLE`. This does duplicate the value from the pungi config file, but I'm not sure it's worth figuring out a bash-y way to parse the value out of the config file just to remove this duplication. And we already effectively did this when discovering `OLDCOMPOSE_ID` and `NEWCOMPOSE_ID`, as the path used there is dependent upon the shortname; to avoid extra duplication, I've changed those to use `$SHORT-$RELEASE_TITLE`. Signed-off-by: Adam Williamson <awilliam@redhat.com>
This repo holds the config files that define what goes in and comes out of a Fedora compose using pungi.