Re: Porting(pulling) a specific changeset from one repository to another
Ahmed A <[email protected]> Wed, 29 Aug 2007 09:46:33 -0700 (PDT)
| Newsgroups | gmane.comp.version-control.bitkeeper.user |
|---|---|
| Message-ID | <[email protected]> |
--- "Eric D. Mudama" <[email protected]> wrote: > On 8/29/07, Ahmed A <[email protected]> wrote: > > Hello, > > > > I am runnning bk version 4.0.2. I have two > repository > > (repo1 and repo2), one is basically a branch of > > another. Changes have been pushed into both repo > over > > time. > > > > I have cloned both of them. I made some changes > to > > repo1, made a changeset (lets call it cs1). I > would > > like to pull ONLY "cs1" into repo2. > > > > I have read the man page for "bk pull", and it > states: > > > > bk pull [-iqRsTu] [-r<rev>] [-E<env>=3D<var>] > [-c<n>] > > [<parent> ...] > > > > -r<rev> Pull up to and including this > revision, > > but exclude later changes. (Or key or changeset > > revision. See bk help terms under 'rev > argument') > > > > I have read the documentation of "rev argument". > > > > Basically, I want to ONLY pull "cs1" from repo1 to > > repo2, and not up to and including "cs1". > > > > Can someone please help me with the syntax? >=20 > This is not an official answer, and I am not > currently a BK user, but > I thought I might be helpful til Larry or someone > else chimes in: >=20 > cs1 in repo1 depends on all the changes that > occurred up to the cs1 in > repo1. All your test data, integration data, etc. >=20 > To pull cs1 by itself, without all those > dependencies, would > effectively make cs1 a fresh delta, which is > basically how BK treats > it. >=20 > You can pull "everything up to a point" into another > repo, but you > cannot cherry pick, because that would give you a > "perceived" > reproducability that doesn't actually exist. >=20 > Ultimately, I think you'll have to export it as a > GNU patch or > equivalent and apply it to repo2. I don't believe > there's any > built-in way to cherrypick a changeset from one repo > to another > without including the ancestor changesets. >=20 > This is a good thing, by the way, but frustrating > until you're used to > working around it. It definitely leads to a very > different workflow > than with most other tools. >=20 > --eric >=20 > PS: Given this restriction on cherry-picking, one > way to solve it is > to base all development for both mainline and the > branch on the > branching changeset. This way all new development > is a child of the > same node, and can be pulled into either mainline or > the branch. >=20 =20 _________________________________________________________________________= ___________ Looking for a deal? Find great prices on flights and hotels with Yahoo! F= areChase. http://farechase.yahoo.com/