[RFC] Multi-parent diffs

[Robert Collins]

On Fri, 2007-04-06 at 14:49 -0500, Matthew D. Fuller wrote:
> 
> It would be useful to be able to store diffs against a revision that's
> not one of the parents 

+1 - There is a TODO about that I think already. Its kindof an
oversight.

> (maybe even one that isn't in the ancestry).

-1. Not being in the ancestry means that we would have to do much more
analysis during 'pull' to be sure we can reconstruct all the texts.

>   For another, it allows a skip-delta-ish approach,
> letting us make fewer reconstruction steps from a fulltext, leading to
> faster construction, or less space from more distance between
> fulltexts, or both.

skip-deltas trade space for time; I think having the flexability to
choose this trade off on a case by case basis is good. See
inventory.knit for instance.

> Actually utilizing such on the fly with commit may prove too
> time-consuming for the most part (though commands like 'revert' and
> 'merge' could drop some hints somewhere for commit to pick up), but
> having the capability latent means that we can later write a
> repack/optimize to use it, or use it in special cases like conversions
> (where we may have more time to do the analysis), or etc. 

Agreed.

-Rob

-- 
GPG key available at: <http://www.robertcollins.net/keys.txt>.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/bazaar/attachments/20070410/ad6e3c76/attachment-0001.pgp