split command should split legacy IDs
If you split a commit with a legacy ID, the legacy ID is copied to both new commits. If used to refer to a commit, it appears that the legacy ID now refers to the second of the split pair. This is problematic if you need to be able to refer in later commands to both new commits independently.
It would be better if the two new commits were given different legacy IDs derived from the original one, for example by appending .split
to the second.
For the record, there is a way to work round this but it is pretty ugly:
<1189> split by webApplication
<1189> setfield legacy_id 1189.split
blob <<EOF
EOF
<1189> inspect
<1189.split> inspect
After the split, the legacy id is present on both commits, but if used as a legacy reference appears to refer to the second part of the split. The legacy id on that can be changed using the setfield
command, the subsequent blob
is required to give that effect by invalidating the legacy ID map.