Do not update project repo when loading results

cadetrdm/batch_running/case.py

@@ -282,6 +282,7 @@ def load(
         allow_commit_hash_mismatch: bool = False,
         allow_options_hash_mismatch: bool = False,
         allow_environment_mismatch: bool = False,
+        fetch: bool = False,
     ) -> Path | None:
         """
         Load results for the current case.
@@ -290,14 +291,13 @@ def load(
             allow_commit_hash_mismatch: If True, allow loading results with mismatched study commit hash.
             allow_options_hash_mismatch: If True, allow loading results with mismatched options hash.
             allow_environment_mismatch: If True, allow loading results with mismatched environment.
+            fetch: If True, fetch output repository refs before looking for matching results.
 
         Returns:
             Path to results.
         """
-        if not self.options.debug:
-            self.project_repo.update()
-        else:
-            print("WARNING: Not updating the repositories while in debug mode.")
+        if fetch:
+            self.output_repo.fetch()
 
         results_branch = self._get_results_branch(
             allow_commit_hash_mismatch=allow_commit_hash_mismatch,

tests/test_read_only_loading.py

@@ -9,7 +9,7 @@
 
 import pytest
 
-from cadetrdm import ProjectRepo, initialize_repo
+from cadetrdm import Case, Options, ProjectRepo, initialize_repo
 from cadetrdm.io_utils import delete_path
 
 
@@ -46,6 +46,25 @@ def repo_with_results(tmp_path):
     return repo
 
 
+@pytest.fixture
+def case_with_results(tmp_path):
+    path_to_repo = tmp_path / "project"
+    initialize_repo(path_to_repo, "results")
+
+    repo = ProjectRepo(path_to_repo)
+    options = Options({"case": "read-only-load"})
+    options.branch_prefix = "read-only-load"
+
+    with repo.track_results(results_commit_message="Add result", options=options):
+        (repo.output_path / "result.csv").write_text("1,2,3\n")
+
+    case = Case(project_repo=repo, options=options, name="read-only-load")
+    cache_folder = repo.cache_folder_for_branch(str(repo.output_repo.active_branch))
+    delete_path(cache_folder)
+
+    return case
+
+
 def test_reading_output_log_leaves_output_repo_untouched(repo_with_results):
     output_repo = repo_with_results.output_repo
     state_before = git_state(output_repo)
@@ -122,3 +141,44 @@ def test_copy_data_to_cache_uses_remote_ref_without_creating_local_branch(repo_w
     assert (cache_path / "result.csv").read_text() == "1,2,3\n"
     assert git_state(output_repo) == state_before
     assert result_branch not in [head.name for head in output_repo._git_repo.heads]
+
+
+def test_case_load_does_not_update_project_repo(case_with_results, monkeypatch):
+    project_repo = case_with_results.project_repo
+    project_state_before = git_state(project_repo)
+    output_state_before = git_state(project_repo.output_repo)
+
+    def fail_update():
+        raise AssertionError("Case.load() updated the project repository")
+
+    def fail_fetch():
+        raise AssertionError("Case.load() fetched output repository refs by default")
+
+    monkeypatch.setattr(project_repo, "update", fail_update)
+    monkeypatch.setattr(project_repo.output_repo, "fetch", fail_fetch)
+
+    results_path = case_with_results.load()
+
+    assert (results_path / "result.csv").read_text() == "1,2,3\n"
+    assert git_state(project_repo) == project_state_before
+    assert git_state(project_repo.output_repo) == output_state_before
+
+
+def test_case_load_can_fetch_output_refs_explicitly(case_with_results, monkeypatch):
+    project_repo = case_with_results.project_repo
+    fetched = False
+
+    def fail_update():
+        raise AssertionError("Case.load() updated the project repository")
+
+    def fetch_output_refs():
+        nonlocal fetched
+        fetched = True
+
+    monkeypatch.setattr(project_repo, "update", fail_update)
+    monkeypatch.setattr(project_repo.output_repo, "fetch", fetch_output_refs)
+
+    results_path = case_with_results.load(fetch=True)
+
+    assert fetched is True
+    assert (results_path / "result.csv").read_text() == "1,2,3\n"