Mike Kupfer wrote: >>>>>> "Rich" == Richard Lowe <richlowe at richlowe.net> writes: >>>>>> > > Mike> What Steve Lau was doing when he ran the deliveries was produce a > Mike> Mercurial bundle. This would contain the changes between the initial > Mike> snapshot and the final as-delivered build. > > Rich> I think they stopped being run before Steve left, but I can't > Rich> remember the reason why, off the top of my head. I'm fairly sure > Rich> there was a reason however, though it may not still apply. > > The email that I have from Steve just says "I've been a little lax in > updating these lately, 66 looks like my most recent one." > > The only issue I know of is that occasionally the respin contains a > one-off fix that conflicts with development on the trunk. That's an > issue for Teamware-based development, too. > Yup. > The answer might be to export the diffs to a text file, but not publish > the bundle, so that somebody doesn't accidentally contaminate their > Mercurial repository with the wrong changeset. >
I'd say publish the bundle anyway - it will just create another head/branch off, but that should still be clean, and users can still update (hg update -C <rev>) to a specific rev in either branch. It'll still be contaminated from the point of view that the changes will be in their repo and might accidentally be pushed - but if there is a pbchk (or incoming gate check) hook that runs to check for the presence of multiple heads, this will prevent it from being pushed to onnv-gate. cheers, steve -- stephen lau | stevel at opensolaris.org | www.whacked.net
