jedcunningham commented on issue #41505:
URL: https://github.com/apache/airflow/issues/41505#issuecomment-2293050972

   This is a better reproduction DAG:
   
   ```
   with DAG("aaa_repro", schedule=None):
       start = EmptyOperator(task_id="start")
   
       a = [
           start
           >> EmptyOperator(task_id=f"a_1_{i}")
           >> EmptyOperator(task_id=f"a_2_{i}")
           >> EmptyOperator(task_id=f"a_3_{i}")
           for i in range(200)
       ]
   
       middle = EmptyOperator(task_id="middle")
   
       b = [
           middle
           >> EmptyOperator(task_id=f"b_1_{i}")
           >> EmptyOperator(task_id=f"b_2_{i}")
           >> EmptyOperator(task_id=f"b_3_{i}")
           for i in range(200)
       ]
   
       middle2 = EmptyOperator(task_id="middle2")
   
       c = [
           middle2
           >> EmptyOperator(task_id=f"c_1_{i}")
           >> EmptyOperator(task_id=f"c_2_{i}")
           >> EmptyOperator(task_id=f"c_3_{i}")
           for i in range(200)
       ]
   
       end = EmptyOperator(task_id="end")
   
       start >> a >> middle >> b >> middle2 >> c >> end
   ```
   
   It uses 5+GB and takes just under 8 minutes to generate on my machine.
   
   And I believe it's less about the whitespace and more that we duplicate 
tasks in the output. For example, this DAG:
   
   ```
   with DAG("aaa_runaway", schedule=None):
       start = EmptyOperator(task_id="start")
       x = [EmptyOperator(task_id=f"x_{i}") for i in range(3)]
       middle = EmptyOperator(task_id="middle")
       y = [EmptyOperator(task_id=f"y_{i}") for i in range(3)]
       end = EmptyOperator(task_id="end")
   
       start >> x >> middle >> y >> end
   ```
   
   which results in this output:
   ```
   <Task(EmptyOperator): start>
       <Task(EmptyOperator): x_0>
           <Task(EmptyOperator): middle>
               <Task(EmptyOperator): y_0>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_1>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_2>
                   <Task(EmptyOperator): end>
       <Task(EmptyOperator): x_1>
           <Task(EmptyOperator): middle>
               <Task(EmptyOperator): y_0>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_1>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_2>
                   <Task(EmptyOperator): end>
       <Task(EmptyOperator): x_2>
           <Task(EmptyOperator): middle>
               <Task(EmptyOperator): y_0>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_1>
                   <Task(EmptyOperator): end>
               <Task(EmptyOperator): y_2>
                   <Task(EmptyOperator): end>
   ```
   
   Note how we get 9 `ends`. If you have a sufficiently complex DAG, this 
format becomes really problematic.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to