On Wed, Dec 14, 2005 at 09:36:54AM -0500, Jason M. Felice wrote:
> I'm wondering if there's a work around for either of the following 
> scenarios:
> 
> I'm using darcs to version some changes to SugarCRM, which is a pretty big
> project in terms of files and data.  SugarCRM just released 4.0.0beta, and
> I've been working on 3.5.1d.  What I wanted to do was this:
> 
> a) Start a new repository.
> b) Pull these patches from my initial repo:
>    b1) Initial import of 3.5.1 (large)
>    b2) 3.5.1d update patch (medium-large)
> c) unzip 4.0.0beta over top
> d) record 4.0.0beta changes
> 
> ... So far, so good ...
> 
> e) Pull my modifications from the initial tree

In my experience this usage is something darcs is not very
good at (but hopefully will do better when the new conflictor
scheme is ready).

The key ingredient for the problem (or at least what may
likely release a problem at some future point) is the
two unsynchronized lines of patch creation, the one where
you develop your own patches and the one where you create
patches from upstream's upgrades.  They don't share context.
The patches created from upstream upgrades will have to
commute with all of your own patches you've ever created, and
if they should conflict, new patches for upstream upgrades will
have to commute, _and conflict_, with these mergers and their
resolutions as well, and to express it with an understatement:
this will not scale.

> No matter what I do, I cannot get e) to complete.  I've tried pushing, 
> pulling, creating a tag.  The first patch that I want to pull from my
> initial tree is very small, and I know it doesn't conflict with anything.

I don't have any guesses for what this problem may be, but
I'd patch in the upstream upgrade on top of my own changes
and record it as "syncing with upstream x.y.z".


-- 
Tommy Pettersson <[EMAIL PROTECTED]>

_______________________________________________
darcs-users mailing list
[email protected]
http://www.abridgegame.org/mailman/listinfo/darcs-users

Reply via email to