Hi list, maybe i'm two trains late with this mail, but if i'm not, here are some ramblings on Ian's bundles-as-repositories idea. To sum it up, i think it's great.
Ian's idea, if i'm not mistaken is that since repositories and bundles are just lists of patches, there is no fundamental difference between darcs apply file.dpatch and darcs pull -a repo. The main difference right now is that you can only apply a bundle if you sent it to that repository. Now look at what happens if you pull from a lazy repository: if they have the patch, then everything works right, else, you will be stuck. But with a bit of imagination, there's something we can do. Imagine (c)lient is pulling patches from (l)azy, but the patches are in fact on (s)erver. If (l) cannot fetch patches from (s) (because it has no space left on disk, or darcs cannot invoke ssh), it could send a "patch thunk" to (c), which has a better chance of being able to fetch the patches for itself. Would that work? So, if you consider a .dpatch as a lazy/partial repository, the analogy is perfect, and with these thunks exchanges, you can apply any bundle to any repository, and darcs magically pulls the missing patches (as long as they are all referenced as being on public repos). Wouldn't it be nice? Florent _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
