[darcs-users] Darcs in kernel-like development model

[Tommy Pettersson]

On Wed, Feb 06, 2008 at 09:01:55AM +0900, Stephen J. Turnbull wrote:
> [...] Also,
> it's not clear to me from following darcs-devel with its two-line
> patches followed by 2000 lines of dependencies that in practice the
> notion of dependency actually used isn't pragmatically equivalent to
> ancestry.)

Just an OT note: the 2000 lines are context, not patch
dependencies.

A brief explanation: A mailed patch by itself in not useful to
darcs, because darcs needs to know in what context the patch
fits. The patch has to be rewritten by darcs to fit in the
context of the target repository. To not have to send the entire
inventory (20 000 lines by now), darcs cuts away the shared part
of the inventory, from the beginning up to the last common
(optimized) tag. This is a very cheap calculation. It would be
possible to calculate a "minimal context", which would contain
only the directly dependent upon patches common between the two
repositories, but that is a _very_ expensive calculation.


-- 
Tommy Pettersson <ptp at lysator.liu.se>