New issue
Advanced search Search tips

Issue 6583 link

Starred by 1 user

Issue metadata

Status: Released
Closed: Apr 2018

Sign in to add a comment

Due to rebase gives false negative on chunk

Project Member Reported by, Jun 27 2017

Issue description

In [1] the removal chunk starting on l.25 of the base (patch 10) should be marked as due_to_rebase, but is not.

Project Member

Comment 1 by, Feb 22 2018

Status: ChangeUnderReview (was: New)
Project Member

Comment 2 by, Feb 26 2018

Labels: -Priority-3 Priority-2
Project Member

Comment 3 by, Apr 6 2018

Labels: -Priority-2 Priority-3
Status: Accepted (was: ChangeUnderReview)
What I fixed in [1] unfortunately doesn't help for the provided example. I had thought that the chunks between the parent of the patch sets and the ones against the bases of the patch sets were only touching each other. I took a closer look and realized that I was wrong. As they overlap, the rebase detection algorithm doesn't even look at them at the moment (which is intended). Improvements for that situation (e.g. splitting up chunks) are only possible after I reworked the diff caches.

Nevertheless, what I fixed with [1] does reduce the number of false negatives in general. So it wasn't in vain.

I'm pumping the priority to P3 again as further improvements necessitate the rework of the diff caches.
Labels: FixedIn-2.15.1
Status: Released (was: Accepted)

Sign in to add a comment