[darcs-users] Replaying patches from a darcs repository

[James D Sadler]

Hi All,

At work we have a very large darcs repository, and due to various
exponential time and space issues,
we want to move to another DVCS.

I have tried converting the repository with tailor and darcs2git.py, and
neither completes: both fail when
darcs runs out of memory.

Essentially, what I want to be able to do is pull the repository one patch
at a time from first to last.
After each pull, I want to be able to capture the changed working tree with
another DVCS (probably git).

That seems to be the mechanism that darcs2git.py implements. However, the
pull on one patch at a time
seems to be what causes darcs to run into exponential time and space issues.


Given that a pull of one-patch-at-a-time is really slow, would it be
possible to craft a darcs repository in such
a way that when I perform the one-patch-at-a-time pull, the repo only
contains patches upto the one I want to pull?
The patches dir (gz files) must only contain patches upto what I need to
pull, and the inventory file must only
reference those patches and no others.

Would this work?  I am concerned because I seem to remember reading
somewhere that darcs retrospectively
modifies the patches during patch commutation, and that would imply that a
subset of the patches in the repo
is not binary identical to when the repo consisted *only* of those patches.
(I could be *way* off base here, so please correct me!).

Thanks for any help,

James.

-- 
James Sadler
james at jamesdsadler.com | james.sadler at mootermedia.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osuosl.org/pipermail/darcs-users/attachments/20071122/97ccc45d/attachment.htm