Re: [PATCH] unpack-trees: fix oneway_merge accidentally carry over stage index

>
> Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:
>
> > One-way merge is supposed to take stat info from the index and
> > everything else from the given tree. This implies stage 0 because trees
> > can't have non-zero stages. The add_entry(.., old, ...) call however
> > will keep stage index from the index.
> >
> > This is normally not a problem if the entry from the index is
> > normal (stage #0). But if there is a conflict, we'll get stage #1 entry
> > as "old" and it gets recorded in the final index. Fix it by clearing
> > stage mask.
> >
> > This bug probably comes from b5b425074e (git-read-tree: make one-way
> > merge also honor the "update" flag, 2005-06-07). Before this commit, we
> > may create the final ("dst") index entry from the one in index, but we
> > do clear CE_STAGEMASK.
>
> Wow, good find.  That's an old one.

> >
> >  t/t2026-checkout-force.sh (new +x) | 26 ++++++++++++++++++++++++++
>
> This makes it cumbersome to have the same fix in the maintenance track
> as t2026 is already in use over there.  Do we need an entirely new test
> just to house this new single test?

> > +test_expect_success 'force checking out a conflict' '
> > +     echo a >a &&
> > +     git add a &&
> > +     git commit -ama &&
> > +     A_OBJ=$(git rev-parse :a) &&
> > +     git branch topic &&
> > +     echo b >a &&
> > +     git commit -amb &&
> > +     B_OBJ=$(git rev-parse :a) &&
> > +     git checkout topic &&
> > +     echo c >a &&
> > +     C_OBJ=$(git hash-object a) &&
> > +     git checkout -m master &&
> > +     test_cmp_rev :1:a $A_OBJ &&
> > +     test_cmp_rev :2:a $B_OBJ &&
> > +     test_cmp_rev :3:a $C_OBJ &&
> > +     git checkout -f topic &&
> > +     test_cmp_rev :a $A_OBJ
>
> So in short, "checkout -f" should have given us an entry for path
> "a", taken from the tip of the 'topic' branch, at stage #0 while
> switching to that branch, but it didn't?  That would be a nice
> summary to have at the beginning of the log message before going
> into the implementation detail of how that happens.