+1 To me the preferred method is to use the context manager (`with DAG(...) as dag:`). We should make sure all examples align with that method if that's not the case already.
On Wed, Jul 3, 2019 at 10:57 AM Kamil Breguła <kamil.breg...@polidea.com> wrote: > This is very confusing. > +1 > > On Wed, Jul 3, 2019 at 7:20 PM Christian Lellmann > <christian.lellm...@googlemail.com.invalid> wrote: > > > Me neither. Also from me +1 (non-binding) on removal. > > > > Tao Feng <fengta...@gmail.com> schrieb am Mi., 3. Juli 2019, 18:38: > > > > > I am not aware of this feature either. And +1 on removing it. > > > > > > On Wed, Jul 3, 2019 at 9:36 AM Kaxil Naik <kaxiln...@gmail.com> wrote: > > > > > > > I am definitely sure that 99% of users, including me, didn't knew > this > > > > feature ever existed 😀. > > > > > > > > It is not a feature worth having tbh. So I am in favor of removing > it. > > > > > > > > Regards, > > > > Kaxil > > > > > > > > On Wed, Jul 3, 2019, 18:37 James Meickle <jmeic...@quantopian.com > > > .invalid> > > > > wrote: > > > > > > > > > I didn't even know this was a feature. Seems like it's > unnecessarily > > > > > ambiguous, since you can't tell at a glance whether a variable is a > > dag > > > > or > > > > > a task. Definitely in favor of removal. > > > > > > > > > > On Wed, Jul 3, 2019 at 8:49 AM Ash Berlin-Taylor <a...@apache.org> > > > wrote: > > > > > > > > > > > I'm just suggesting removing the `dag >> task` -- `task >> task` > > will > > > > > stay > > > > > > > > > > > > > On 3 Jul 2019, at 13:46, Philippe Gagnon < > philgagn...@gmail.com> > > > > > wrote: > > > > > > > > > > > > > > Just to be clear, are you suggesting removing all bitshift > > operator > > > > > > > overloads from airflow operators (sorry - the dual meaning of > > > > operator > > > > > > here > > > > > > > is confusing), or just the assignment to DAG behavior? > > > > > > > > > > > > > > If it's the former, I find it to be a particularly expressive > way > > > to > > > > > > define > > > > > > > dependencies between tasks so I would vote to keep it as is. > The > > > > latter > > > > > > > usage is much less useful, so I would be +1 on removing it. > > > > > > > > > > > > > > On Wed, Jul 3, 2019 at 8:42 AM Ash Berlin-Taylor < > a...@apache.org > > > > > > > > wrote: > > > > > > > > > > > > > >> It is possible to assign a task to the dag using the bitshift > > > > > operators, > > > > > > >> however it doesn't pick up default_args when done this way < > > > > > > >> https://issues.apache.org/jira/browse/AIRFLOW-883>: > > > > > > >> > > > > > > >> ``` > > > > > > >> dag = DAG('my_dag', default_args=default_args) > > > > > > >> dummy = DummyOperator(task_id='dummy') > > > > > > >> > > > > > > >> dag >> dummy > > > > > > >> ``` > > > > > > >> > > > > > > >> We could fix that, but how about instead we remove this way of > > > > > assigning > > > > > > >> tasks to dags, leaving the context manager (`with dag:`) and > > other > > > > > > >> constructions (`Operator(..., dag=dag)`) > > > > > > >> > > > > > > >> Thoughts? > > > > > > >> > > > > > > >> > > > > > > > > > > > > > > > > > > > > > > > > > > >
