add/commit: figure out whether to link or not

dvc/output.py

@@ -542,7 +542,8 @@ def _build(
             desc=f"Collecting files and computing hashes in {self}",
             disable=no_progress_bar,
         ) as pb:
-            return build(*args, callback=pb.as_callback(), **kwargs)
+            kwargs["callback"] = pb.as_callback()
+            return build(*args, **kwargs)
 
     def _get_hash_meta(self):
         if self.use_cache:
@@ -751,9 +752,8 @@ def commit(self, filter_info=None, relink=True) -> None:
             )
             # NOTE: trying to use hardlink during transfer only if we will be
             # relinking later
-            hardlink = relink
             if granular:
-                obj = self._commit_granular_dir(filter_info, hardlink)
+                obj = self._commit_granular_dir(filter_info, hardlink=False)
             else:
                 staging, _, obj = self._build(
                     self.cache,
@@ -771,7 +771,7 @@ def commit(self, filter_info=None, relink=True) -> None:
                         self.cache,
                         {obj.hash_info},
                         shallow=False,
-                        hardlink=hardlink,
+                        hardlink=False,
                         callback=cb,
                     )
             if relink:
@@ -1403,7 +1403,7 @@ def add(  # noqa: C901
                 staging,
                 self.cache,
                 {obj.hash_info},
-                hardlink=relink,
+                hardlink=False,
                 shallow=False,
                 callback=cb,
             )
@@ -1421,6 +1421,7 @@ def add(  # noqa: C901
                     state=self.repo.state,
                     prompt=prompt.confirm,
                     progress_callback=callback,
+                    old=obj,
                 )
             self.set_exec()
         return obj

pyproject.toml

@@ -38,7 +38,7 @@ dependencies = [
     "distro>=1.3",
     "dpath<3,>=2.1.0",
     "dulwich",
-    "dvc-data>=3.15,<3.16",
+    "dvc-data @ git+https://github.com/iterative/dvc-data.git@relinking-diff",
     "dvc-http>=2.29.0",
     "dvc-objects",
     "dvc-render>=1.0.1,<2",

tests/func/test_add.py

@@ -305,20 +305,20 @@ def test_should_update_state_entry_for_directory_after_add(mocker, dvc, tmp_dir)
 
     ret = main(["add", "data"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 4
+    assert file_md5_counter.mock.call_count == 3
 
     ret = main(["status"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 5
+    assert file_md5_counter.mock.call_count == 4
 
     os.rename("data", "data.back")
     ret = main(["checkout"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 6
+    assert file_md5_counter.mock.call_count == 5
 
     ret = main(["status"])
     assert ret == 0
-    assert file_md5_counter.mock.call_count == 7
+    assert file_md5_counter.mock.call_count == 6
 
 
 def test_add_commit(tmp_dir, dvc):
@@ -339,15 +339,15 @@ def test_should_collect_dir_cache_only_once(mocker, tmp_dir, dvc):
     counter = mocker.spy(dvc_data.hashfile.build, "_build_tree")
     ret = main(["add", "data"])
     assert ret == 0
-    assert counter.mock.call_count == 2
+    assert counter.mock.call_count == 1
 
     ret = main(["status"])
     assert ret == 0
-    assert counter.mock.call_count == 3
+    assert counter.mock.call_count == 2
 
     ret = main(["status"])
     assert ret == 0
-    assert counter.mock.call_count == 4
+    assert counter.mock.call_count == 3
 
 
 def test_should_place_stage_in_data_dir_if_repository_below_symlink(
@@ -754,6 +754,7 @@ def test_add_file_in_symlink_dir(make_tmp_dir, tmp_dir, dvc):
 def test_add_with_cache_link_error(tmp_dir, dvc, mocker, capsys):
     tmp_dir.gen("foo", "foo")
 
+    dvc.cache.local.cache_types = ["symlink", "hardlink"]
     mocker.patch("dvc_data.hashfile.checkout.test_links", return_value=[])
     dvc.add("foo")
     err = capsys.readouterr()[1]