get/import: use in-memory index for remote repositories

dvc/repo/__init__.py

@@ -232,6 +232,7 @@ def __init__(  # noqa: PLR0915, PLR0913
             Callable[[str, Exception], None]
         ] = None
         self._lock_depth: int = 0
+        self._is_remote = False
 
     def __str__(self):
         return self.url or self.root_dir
@@ -365,6 +366,8 @@ def data_index(self) -> "DataIndex":
             index_dir = os.path.join(self.site_cache_dir, "index", "data")
             os.makedirs(index_dir, exist_ok=True)
             self._data_index = DataIndex.open(os.path.join(index_dir, "db.db"))
+        if self._is_remote:
+            self._data_index = DataIndex()
 
         return self._data_index
 

dvc/repo/open_repo.py

@@ -40,7 +40,9 @@ def _external_repo(url, rev: Optional[str] = None, **kwargs) -> "Repo":
         **kwargs,
     )
 
-    return Repo(**repo_kwargs)
+    repo = Repo(**repo_kwargs)
+    repo._is_remote = True
+    return repo
 
 
 def open_repo(url, *args, **kwargs):

dvc/testing/benchmarks/cli/commands/test_get.py

@@ -6,4 +6,4 @@ def test_get(bench_dvc, tmp_dir, scm, dvc, make_dataset, remote):
     dataset = make_dataset(
         cache=False, files=False, dvcfile=True, commit=True, remote=True
     )
-    bench_dvc("get", tmp_dir, dataset.name, "-o", "new")
+    bench_dvc("get", f"file://{tmp_dir.as_posix()}", dataset.name, "-o", "new")

dvc/testing/benchmarks/cli/commands/test_import.py

@@ -1,7 +1,6 @@
 import pytest
 
 
-@pytest.mark.flaky(reruns=3)
 @pytest.mark.requires(
     "!=3.53.*,!=3.54.0",
     reason="Takes 10 mins to run. Regression in 3.53.0, fixed in 3.54.1",
@@ -12,4 +11,4 @@ def test_import(bench_dvc, tmp_dir, scm, dvc, make_dataset, remote):
     dataset = make_dataset(
         cache=False, files=False, dvcfile=True, commit=True, remote=True
     )
-    bench_dvc("import", tmp_dir, dataset.name, "-o", "new")
+    bench_dvc("import", f"file://{tmp_dir.as_posix()}", dataset.name, "-o", "new")