> On Thu, Apr 25, 2013 at 1:23 PM, Marcel Reutegger <[email protected]> > wrote: > > I'd be interested to hear what people think about this problem. > > should we retry commits when these kind of conflicts where > > introduced by hooks or editor or do we require the latter to > > take care of this and only manipulate content in a non-conflicting > > way? > > Note that the SegmentMK is designed to avoid this problem; it can > guarantee that the commit hooks get run against the latest state of > the repository and thus can't possibly conflict.
it does it with a retry logic, right? the same could be implemented in the KernelNodeStore... > More generally though the best approach here is probably to combine > the approaches you mention. Commit hooks should generally try to avoid > creating conflicts (like what the p2 index is already doing), but if > doing so is not possible or too expensive (as it in many cases may be) > it would be best if the commit could fail and be retried after > rebasing. sounds like a good plan. regards marcel
