| 2021-03-08 19:56:58 |
Bryce Harrington |
description |
`git ubuntu merge finish` will construct the debian/changelog entry for merges. One workflow style used by the team is to create empty commits for dropped changes, so that entries get created for them in the changelog. These commits should then be added under a " * Dropped:" section at the end of the changelog.
This can be achieved manually by rebasing the git history and move all the dropped commits to the end, with the first of this set given a commit message that starts with " * Dropped:". This is an effective workaround, but has the downside that it loses the historical ordering, and it requires some care to make sure the commits remain in proper order.
A different convention, used by those who don't rely on the automatic debian/changelog generation, is to mark all the dropped commits with " * Dropped:" or similar, and then using them to inform the manual writing of the changelog entry. So, one approach might be to simply scan for commits beginning with this string (and similar ones) and when generating the changelog, strip that out and gather those commits at the end after a " * Dropped:" label. |
`git ubuntu merge finish` will run reconstruct-changelog to generate the debian/changelog entry for merges. One workflow style used by the team is to create empty commits for dropped changes, so that entries get created for them in the changelog. These commits should then be added under a " * Dropped:" section at the end of the changelog.
This can be achieved manually by rebasing the git history and move all the dropped commits to the end, with the first of this set given a commit message that starts with " * Dropped:". This is an effective workaround, but has the downside that it loses the historical ordering, and it requires some care to make sure the commits remain in proper order.
A different convention, used by those who don't rely on the automatic debian/changelog generation, is to mark all the dropped commits with " * Dropped:" or similar, and then using them to inform the manual writing of the changelog entry. So, one approach might be to simply scan for commits beginning with this string (and similar ones) and when generating the changelog, strip that out and gather those commits at the end after a " * Dropped:" label. |
|
