Timeout on status traces that never finish

CHANGELOG.md

@@ -10,6 +10,7 @@
 * New option `--verbose` for the `datasets ls` command which lists every
   relevant metadata field for the listed datasets.
 * New output format option for printing CSV: `--format=csv`.
+* Better error messages for status traces that never finish in `status --watch`.
 * Improved SIGPIPE handling so that `okdata` works better with tools
   such as `head`.
 

okdata/cli/commands/status.py

@@ -101,13 +101,23 @@ def wait_until_done(self, trace_id):
         if trace_status != "FINISHED":
             print("Waiting for processing to finish", end="", flush=True)
 
-        while trace_status != "FINISHED":
+        attempts = 0
+        max_attempts = 100
+        while trace_status != "FINISHED" and attempts < max_attempts:
             print(".", end="", flush=True)
             sleep(2)
+            attempts += 1
             trace_events = self.get_trace_events(trace_id)
             trace_status = StatusCommand.find_latest_event(trace_events)["trace_status"]
         print()
 
+        if attempts == max_attempts:
+            print(
+                "\nWe've waited for a good while now, and processing hasn't "
+                "yet finished.\nSomething else is most likely wrong, maybe a "
+                "pipeline hasn't been configured for the dataset?\n"
+            )
+
         return trace_events
 
     def status_for_id(self, trace_id):