>>>>> "Ken" == Ken Erickson <ken.erickson at sun.com> writes:
Ken> I guess what I am wondering is, are we planning to use hooks to Ken> closely mimic the current behavior of teamware, or is someone Ken> re-inventing something else? Or is this part of what I should be Ken> doing as I work on these gate scripts? I think that part of the work on the gate scripts is to make that decision--do we emulate the current Teamware-based approach, or do we do something that fits more naturally with Mercurial. Factors to consider would include - engineering effort - lock-in (e.g., can we do something simple but slow to get started, then improve later) - flexibility (e.g., for correcting problems) Talking about buglist specifically, we would lose some flexibility by getting the list from Mercurial (as opposed to getting it from a mail log). For example, with the mail log approach, we can edit the mail log to correct bogus putback messages. On the other hand, the plan for the gate hooks is that they will detect things like bogus putback comments and prevent such changesets from ever getting into the gate in the first place. If the hooks work as intended, great. If there's a bug and they let in bad comments, it'll mean more work for the gatekeeper to produce the bug list. I don't know what the right answer is here. Maybe the thing to do is make a tentative plan and run it by various groups of people. Recent gatekeepers are still around and will know what sorts of issues they tended to run into. A lot of the things that the really early gatekeepers ran into will no longer apply, because the whole gate model has changed since then (thanks to the introduction of the clone workspace in 5.5). mike
