Danek Duvall wrote:
On Thu, Dec 07, 2006 at 04:28:22PM -0800, Stephen Lau wrote:How do we want to model the bugfix/putback policy for ON? I see a few options: 1) 1 changegroup == 1 changeset == multiple bugs This is basically what we have now... multiple bugs fixed, integrated as one changeset, with one changeset per putback/changegroup i.e.: prior to doing a 'hg push', all commits are redelget'd into one 2) 1 changegroup == multiple changesets, 1 bug/changeset This is a more rigorous (IMHO) option where each bugfix is done as one 'hg commit', with the actual putback (nay, changegroup) being multiple commitsI think in the long run, I'd prefer the second option. As James points out, it makes it far easier to tease out what files belong to what bugfixes. However, it's going to make larger wads more difficult to deal with until people get fluent with mq (which is the only thing that makes such a thing possible at all). It's probably worth writing deltachk so that it can support either model, and have a per-repo switch to select the model (or something where #2 goes through without comment, and #1 sends back a message saying that #2 is preferred).
A per-repo switch is an interesting idea.... maybe for consolidations that prefer less rigorous tracking. That's something we can investigate.
And yes, this will definitely require the use of Mq; which at this point I'm starting to assume is a given. There will have to be some additional work, either at the Mq level, or at our mythical Hg-wx-whizzy-extension level to wrap some of the Mq functions to make it easier for people more used to a simple 'wx redelget' to use.
cheers, steve -- stephen lau // [EMAIL PROTECTED] | 650.786.0845 | http://whacked.net opensolaris // solaris kernel development _______________________________________________ tools-discuss mailing list tools-discuss@opensolaris.org
