[darcs-users] How to pull all my patches from one repository into another?
Quag
quaggy at gmail.com
Fri Jan 21 04:47:29 UTC 2005
On Thu, 20 Jan 2005 19:02:04 -0800, Alexey N. Solofnenko
<A.Solofnenko at mdl.com> wrote:
> At first I created one base repository, then created a branch with my
> changes. Now I created a better base repository and I want create
> another branch from that repository with my changes. How can I pull all
> my changes from one repository into another?
Hi Alexey,
I don't think you can. This is a fundamental limitation with darcs.
This limitation usually shows up when a number of people start
tracking a non-darcs-based project. Each person creates their own
"base" repository and is unable to share patches with the others using
darcs.
I think Zooko was the first to raise this issue. For more information
try searching the list archives for his posts.
A while ago I pondered a solution, but couldn't pin anything down. I
was trying to mark patches as either "original content" (something you
have personally created) or "published" (for patches created from a
non-darcs public source). The merger rules for "original content"
would be the same as now, but new merger rules would have to be
created for the "published" patches. Also a new set of merger rules
have to be created for the interactions between "published" and
"original content" patches.
All of the ideas I came up with for dealing with merges involving
"published" patches required that both patches were present on the
local machine, or for the user to just assert that the patches are the
same. Having to fetch both "published" patches becomes a problem when
dealing with large source trees (like the linux kernel...).
Currently when using diff, patch and tar the user just tells patch
that the underlying "published" patches (the tar balls) are the same
on both ends. The user uses meta information (like people saying which
kernel version the patch is for) to decide if the underlying working
trees were the same or not. My ideas ended up relying on the same kind
of user-based assertion.
Jonathan Wright.
More information about the darcs-users
mailing list