http://bugs.grommit.com/show_bug.cgi?id=407
Summary: cdm's branch handling may still confuse people Product: SCM Migration Version: unspecified Platform: All OS/Version: Solaris 11/Nevada Status: NEW Severity: normal Priority: P3 Component: cdm AssignedTo: scm-migration-dev at opensolaris.org ReportedBy: richlowe at richlowe.net I'm filing this as a catch-all for the branching bits that behave somewhat surprisingly, either such that they can be fixed, or documented (or both). 1. We pretend the dirstate is the head of a branch, making the active list somewhat confusing if you're at a non-head changeset. (it is truncated at the current working revision) 2. We do not warn the user if they are in a situation where #1, above, will occur. 3. Backup treats the dirstate as if it's always a branch head, when deciding if your working copy changed. 4. We may assume in various places that in a merge, the remote parent is always the second. This isn't true. The root cause is primarily technical, in that if you are back in workspace history, there is the possibility of the branch you are on having more than one head, and no obvious way to guess which head is the one the user would intend or whether, in fact, the working copy *should* be treated as a new head (perhaps they're working back in the history such as to create a new branch, for instance). The options I see (for the first two in that list) are to error if they aren't at a head (where a modified working copy is taken as a head wherever it may occur), to guess a suitable head, perhaps the tip-most head of the current branch (which would most recent on that branch). I suspect that in any case, we should warn the user that we're taking a guess, or treating any revision other than their current revision as the head. -- Configure bugmail: http://bugs.grommit.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug, or are watching the assignee.