chore(experiment-runner): ensure backwards compatibility for old server versions

Author: marliessophie

langfuse/_client/client.py

@@ -130,6 +130,27 @@
 from langfuse.types import MaskFunction, ScoreDataType, SpanLevel, TraceContext
 
 
+def _is_unrecognized_field_error(e: Exception, field_name: str) -> bool:
+    """Check if error contains an unrecognized field.
+
+    Used for backward compatibility when newer SDK sends fields not supported by older servers.
+
+    Args:
+        e: Exception from API call
+        field_name: Name of field to check for in error
+
+    Returns:
+        True if error indicates field_name is unrecognized by the server
+    """
+    try:
+        return any(
+            field_name in error.get("keys", [])
+            for error in getattr(e, "body", {}).get("error", [])
+        )
+    except (AttributeError, TypeError):
+        return False
+
+
 class Langfuse:
     """Main client for Langfuse tracing and platform features.
 
@@ -2937,7 +2958,35 @@ async def _process_experiment_item(
                         dataset_run_id = dataset_run_item.dataset_run_id
 
                     except Exception as e:
-                        langfuse_logger.error(f"Failed to create dataset run item: {e}")
+                        # Only retry if the error is specifically about datasetVersion being unrecognized
+                        # This handles backward compatibility with older server versions
+                        if _is_unrecognized_field_error(e, "datasetVersion"):
+                            langfuse_logger.warning(
+                                "Server doesn't support datasetVersion field, retrying without it"
+                            )
+                            try:
+                                dataset_run_item = await asyncio.to_thread(
+                                    self.api.dataset_run_items.create,
+                                    request=CreateDatasetRunItemRequest(
+                                        runName=experiment_run_name,
+                                        runDescription=experiment_description,
+                                        metadata=experiment_metadata,
+                                        datasetItemId=item.id,  # type: ignore
+                                        traceId=trace_id,
+                                        observationId=span.id,
+                                        # Note: datasetVersion omitted for backward compatibility
+                                    ),
+                                )
+                                dataset_run_id = dataset_run_item.dataset_run_id
+                            except Exception as retry_error:
+                                error_msg = (
+                                    f"Failed to create dataset run item: {retry_error}"
+                                )
+                                langfuse_logger.error(error_msg)
+                        else:
+                            # Different error, just log it
+                            error_msg = f"Failed to create dataset run item: {e}"
+                            langfuse_logger.error(error_msg)
 
                 if (
                     not isinstance(item, dict)