RE: Maintaining 'stable' and 'development' versionsof a repository (hopeless newbie question)
"Andy Chittenden" <[email protected]> Mon, 20 Jun 2005 15:54:01 +0100
| Newsgroups | gmane.comp.version-control.bitkeeper.user |
|---|---|
| Message-ID | <89E85E0168AD994693B574C80EDB9C27022BA465@uk-email.terastack.bluearc.com> |
> I think you can do a limited form of "cherry-picking" by exporting > changeset D from the development branch, to a patchfile, and then > importing that patchfile into the stable branch. That works well as far as it goes. However, there are a few drawbacks: o you have to reenter the change history as you're not using BK to achieve this o what if the "cherry-picked" patch relies on a small fragment of one of the "rejected" changesets? You'd then need to either reapply a portion of one of the patchs for those files affected as well or change the patch you're applying - I guess patch would generate .rej files in this case so you knew you had to do something. o both these get worse and worse as the number of changed files in each of the ChangeSets grows and grows. o if one the files being "cherry-picked" is new to the stable version, you end up in a backporting nightmare when you pull through the updated stable version into your experimental repos: see <http://db.bitkeeper.com/cgi-bin/view.cgi?.page=view&id=2002-07-17-007> -- Andy, BlueArc Engineering