This is a problem for two reasons. First, removing changesets is done by simply editing the file. You have to remember which changesets need to be removed by their hash. It's doable, but could be more user-friendly here.
Second, and more seriously, upstream repositories may rebase out these changes. If the local ./src cache has been removed or the repo recently cloned, there may not be enough information to determine what a changeset used to refer to. If we keep some basic information about the changeset around, we can at least let a user know what is missing, and maybe even find it automatically for them.
This is a problem for two reasons. First, removing changesets is done by simply editing the file. You have to remember which changesets need to be removed by their hash. It's doable, but could be more user-friendly here.
Second, and more seriously, upstream repositories may rebase out these changes. If the local
./srccache has been removed or the repo recently cloned, there may not be enough information to determine what a changeset used to refer to. If we keep some basic information about the changeset around, we can at least let a user know what is missing, and maybe even find it automatically for them.