Hi, Thomas Moschny wrote:
There's one corner case to be taken into account. Consider a branch with two heads, and let's say they can be merged without conflicts (or the conflicts can be solved in a straightforward manner). Now two developers merge them independently. They end up with exactly the same revision (with the very same hash), and when they sync later, we get one revision with two author certs, two dates, and two changelogs. A bit strange at first sight, but meaningful. Now, consider this with author, date, and changelog being part of the revision (and thus the hash): The results of the two developers doing the merge will not be the same, due to their different author names, and (likely) due to their different commit date. We end up with two 'merged' revisions, and the branch again has two heads. After syncing, the devs have to do a merge again, ... and again, ... :)
Yes, good point. So we want date, changelog and author to be written to an existing revision.
[ Or else we'd have to make sure that such a merge would somehow resolve automatically into one revision with both authors and an avg. (or max) date. Also does not seem to be trivial, given that we don't really gain that much... ]
Regards Markus _______________________________________________ Monotone-devel mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/monotone-devel
