Markreleases are not done for stable releases, causing the log messages for the releases to be merged under a single release in the rpm changelog.
e.g. if there are foo-1-1.1.mga1, foo-1-1.2.mga1, foo-1-1.3.mga1 in the updates repository, one gets
* Tue Oct 04 2011 foobar 11-1.3.mga1
- log messages of 1.1
- log messages of 1.2
- log messages of 1.3
+ baz <baz>
- some other packager's log message for 1.2
+ foobaz <foobaz>
- some third packager's log message for 1.3
Either the youri markrelease action should run markrelease in the originating tree
instead of cauldron, or add making markrelease to the final update move (from
_testing to updates) script.
The former approach would probably be better.
Still valid today.