The ability to record and make use of the linear sequence of states a repository has been in.
For example:
post-commit and post-incoming trigger which take some index, add 1 and make a tag ${BRANCH}_$INDEX .
Only have that trigger on master repos, to save from everyone propagating their personal event stack around.
Note that a node could have more than one tag on it: at some point bugfix becomes dev so the tip would have both bugfix.xxx and dev.zzz .
Alternative from Wayne:
Just calling creating a new dev tag every time the tip of -dev changes would do the same thing. The previous values of dev are still recorded as the old bk-tag-del RTI shows.
Queries
When did this bugfix cset make it into dev? bk set -tt -l -r${BUGFIX}_$INDEX | grep dev_[_]*$ | tail -1
Bisect going only over this repo’s event stack. I.e. don’t jump around branches, just test the merges and print the branch this was merged from.