Noam Postavsky writes: >> I think the warning should be improved: >> - I shouldn't get a warning in the above case, since this commit doesn't >> *add* those files (they weren't on origin/master admittedly but they were >> already on HEAD). > > We could choose which side of the merge to check based on an > environment var (that was considered in the thread I referenced above, > but we didn't see much of a use case at the time). Or is it possible > to check only changes from the merge itself (i.e., in case of conflict > resolution)? > >> - the warning should give me some hint about which file fails the test. > > Yeah, that would make sense. This seems fairly difficult to do while keeping in standard bourne shell syntax. Not sure how to catch file names with whitespace.