pruning history from branch - rebase?

[David Reitter]

On Mar 20, 2009, at 7:24 PM, John Arbash Meinel wrote:
>
> bzr init
> bzr commit -m "initial null revision"
> bzr merge -r0..-1 ../big-project
> bzr revert --forget-merges
> bzr commit
>
> Will merge the current content of the files, and not the actual
> ancestry.

OK, that seems easy.  If I just merge in the first 80k revs, the do -- 
forget-merges, and then merge in the remaining ones, then I should  
have a partial history, right?
[I'm trying that right now, but it seems to take a while.  Progress  
bar doesn't seem to be changing.. ]

Merging the remainder (with history) se would require me to merge in  
the upstream:

> Note that the *reverse* is not 100% true. Specifically, tiny-project  
> has
> none of the ancestry of big-project. If you want to merge from
> big-project back into tiny-project, you have to use a 'cherry-pick' to
> do so (assuming you don't want to actually ever bring in the rest of  
> the
> ancestry of big-project, which I assume was the whole point of the
> exercise).

By "cherry-pick", you mean that I just have to merge with a -r  
argument (doing -r80000 or something like that), right?
That is easy enough.  I'd like to merge in upstream changes very  
regularly, perhaps even every night and automatically unless there are  
conflicts.

Basically I want to keep this branch in a repository on a server so  
that our downstream project just deals with the latest couple of years  
of revisions.  I want to have _some_ revisions though rather than  
none, since they are quite informative.

- Davdi

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2193 bytes
Desc: not available
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20090320/1915e07b/attachment-0001.bin