Components that are no longer needed or are not needed by a subset of users, can be deleted.

Note: this is partially implemented, and these notes are to help anyone who is completing the work, to not have to think of all the scenarios from scratch.

When a repo has a deleted component still populated, then it is still visible. When it has a deleted component that is missing, then it is invisible. This is both a plus and a minus.


Clone works by cloning the product, then using the remote HERE file to populate. If the remote HERE file is ALL, and the clone destination only has the product, then the deleted component will be hidden, and won’t be included in the clone.

Better would be if a clone of a repo with the deleted component still present would clone that component. Otherwise, that component is vulnerable to getting lost.


Two problems scenarios with pull: merge and update.


bk clone . ../client1
bk rmcomp x
bk clone . ../no-x
cd client1/x
# do work; make csets
cd ../../no-x
pull ../client1

This fails because component x needs a merge, but is not populated.

A solution is to have client pulled into a copy of the main repo that has the deleted component populated. The merge is done there, then can be pulled into no-x.


cd main
bk rmcomp x
bk clone . ../no-x
cd client1/x
bk unrmcomp x
# do work; make csets
cd ../no-x
pull ../main

This fails because the the product is pulled which re-activates x but x isn’t populated and an error is generated.

The solution is to have x look like a new component in this case, and get fed to the populate code.