filter-branch: fix ref rewriting with --subdirectory-filter

The previous ancestor discovery code failed on any refs that are (pre-rewrite) ancestors of commits marked for rewriting. This means that in a situation A -- B(topic) -- C(master) where B is dropped by --subdirectory-filter pruning, the 'topic' was not moved up to A as intended, but left unrewritten because we asked about 'git rev-list ^master topic', which does not return anything. Instead, we use the straightforward git rev-list -1 $ref -- $filter_subdir to find the right ancestor. To justify this, note that the nearest ancestor is unique: We use the output of git rev-list --parents -- $filter_subdir to rewrite commits in the first pass, before any ref rewriting. If B is a non-merge commit, the only candidate is its parent. If it is a merge, there are two cases: - All sides of the merge bring the same subdirectory contents. Then rev-list already pruned away the merge in favour for just one of its parents, so there is only one candidate. - Some merge sides, or the merge outcome, differ. Then the merge is not pruned and can be rewritten directly. So it is always safe to use rev-list -1. Signed-off-by: Thomas Rast <trast@student.ethz.ch> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Thomas Rast committed Aug 12, 2008 at 10:45 UTC a0e46390d397e71182e42930b98b6b59a1a84898

2 files changed +12 -17

git-filter-branch.sh

+11 -16

@@ -317,24 +317,19 @@ done <../revs

317

318 # In case of a subdirectory filter, it is possible that a specified head

319 # is not in the set of rewritten commits, because it was pruned by the

320 -# revision walker. Fix it by mapping these heads to the next rewritten

321 -# ancestor(s), i.e. the boundaries in the set of rewritten commits.

320 +# revision walker. Fix it by mapping these heads to the unique nearest

321 +# ancestor that survived the pruning.

322

323 -# NEEDSWORK: we should sort the unmapped refs topologically first

324 -while read ref

325 -do

326 - sha1=$(git rev-parse "$ref"^0)

327 - test -f "$workdir"/../map/$sha1 && continue

328 - # Assign the boundarie(s) in the set of rewritten commits

329 - # as the replacement commit(s).

330 - # (This would look a bit nicer if --not --stdin worked.)

331 - for p in $( (cd "$workdir"/../map; ls | sed "s/^/^/") |

332 - git rev-list $ref --boundary --stdin |

333 - sed -n "s/^-//p")

323 +if test "$filter_subdir"

324 +then

325 + while read ref

326 do

335 - map $p >> "$workdir"/../map/$sha1

336 - done

337 -done < "$tempdir"/heads

327 + sha1=$(git rev-parse "$ref"^0)

328 + test -f "$workdir"/../map/$sha1 && continue

329 + ancestor=$(git rev-list -1 $ref -- "$filter_subdir")

330 + test "$ancestor" && echo $(map $ancestor) >> "$workdir"/../map/$sha1

331 + done < "$tempdir"/heads

332 +fi

333

334 # Finally update the refs

335

t/t7003-filter-branch.sh

+1 -1

@@ -101,7 +101,7 @@ test_expect_success 'filter subdirectory only' '

101 refs/heads/sub refs/heads/sub-earlier

102 '

103

104 -test_expect_failure 'subdirectory filter result looks okay' '

104 +test_expect_success 'subdirectory filter result looks okay' '

105 test 2 = $(git rev-list sub | wc -l) &&

106 git show sub:new &&

107 test_must_fail git show sub:subdir &&