Ahha, I was wondering if there was something like typing.TYPE_CHECKING but for static analyzers/IDEs. I like that approach. Testing it out now.
On Feb 23 2020, at 10:28 pm, Kamil Breguła <kamil.breg...@polidea.com> wrote: > Hello, > > I admit that I do not follow the subject closely, but I think that > here is the answer. > https://github.com/celery/celery/blob/master/celery/__init__.py#L63-L78 > > We can also create a module with lazy-loaded attributes: > https://github.com/celery/celery/blob/master/celery/__init__.py#L158-L184 > > Best regards, > Kamil > > On Sun, Feb 23, 2020 at 11:20 PM Ash Berlin-Taylor <a...@apache.org> wrote: > > > > I've managed to get type stubs working in VSCode (and intellij), so that > > may be an option. > > TBH I think I'm leaning somewhat towards 5 -- after all that would then go > > some way to clarify the distinction between DAG and DagModel (i.e. DAG > > doesn't belong in models, it's not a DB model) > > On Feb 23 2020, at 10:15 pm, Kaxil Naik <kaxiln...@gmail.com> wrote: > > > 1. Just have a deprecation warning (which it's not even clear if PEP-562 > > > style will show a nice (!) warning in IDEs or just at run time > > > 2. I learn to live with from airflow.models.dag import DAG > > > 3. After we change airflow.models to not import all the sub-modules > > > automatically, go back to importing "airflow.models.dag.DAG" into > > > "airflow". > > > 4. We create a new (abstract) base class that we _can_ import in to the > > > airflow module that airflow.models.dag.DAG extends. This could be > > > automated > > > by a pre-commit so it is always up to date with the real DAG class > > > 5. We create a new module of things that are part of the public/dag > > > author API, such as airflow.dag (currently just has a "BaseDAG" sub-module > > > but that is barely used, and only by the scheduler so could be removed) > > > > > > > > > Happy with (3) and/or (5) > > > (1) does not work with Pycharm - it doesn't show deprecation warning. > > > ᐧ > > > > > > On Sun, Feb 23, 2020 at 10:10 PM Ash Berlin-Taylor <a...@apache.org> > > > wrote: > > > > Gah not only that, vscode also ignores __all__ settings which is really > > > > bad form. > > > > > > > > Pycharm at least does respect __all__ for what to show in the module, > > > > and > > > > also respects type stub files (.pyi) when completing constructor args > > > > etc. > > > > In short IDEs are terrible at dynamic languages, but it is what it is. > > > > So a few options then: > > > > Just have a deprecation warning (which it's not even clear if PEP-562 > > > > style will show a nice (!) warning in IDEs or just at run time > > > > I learn to live with from airflow.models.dag import DAG > > > > > > > > After we change airflow.models to not import all the sub-modules > > > > automatically, go back to importing "airflow.models.dag.DAG" into > > > > "airflow". > > > > > > > > We create a new (abstract) base class that we _can_ import in to the > > > > airflow module that airflow.models.dag.DAG extends. This could be > > > > automated > > > > by a pre-commit so it is always up to date with the real DAG class > > > > > > > > We create a new module of things that are part of the public/dag author > > > > API, such as airflow.dag (currently just has a "BaseDAG" sub-module but > > > > that is barely used, and only by the scheduler so could be removed) > > > > > > > > Anyone have any strong opinions? I think I'd favour 3, 4 or 5 > > > > -ash > > > > On Feb 22 2020, at 10:50 pm, Kaxil Naik <kaxiln...@gmail.com> wrote: > > > > > I tried an example with PEP-562 but the autocomplete didn't work in > > > > > > > > > > > > Pycharm > > > > > and still showed the deprecated function. > > > > > > > > > > I agree we should have a deprecation warning before we should have > > > > changed > > > > > it. How about we introduce a deprecation warning in the next version > > > > > (1.10.10) ? > > > > > > > > > > Regards, > > > > > Kaxil > > > > > ᐧ > > > > > > > > > > On Sat, Feb 22, 2020 at 5:35 PM Jarek Potiuk > > > > > <jarek.pot...@polidea.com> > > > > > wrote: > > > > > > > > > > > I also started to use VSCode in parallel on my Chromebook as this is > > > > the > > > > > > best way to get devenv running there (and I believe it's now THE > > > > > > most > > > > > > popular IDE - including for Python developers). I can check it > > > > > > there as > > > > > > well. > > > > > > > > > > > > On Sat, Feb 22, 2020 at 6:33 PM Jarek Potiuk > > > > > > <jarek.pot...@polidea.com > > > > > > wrote: > > > > > > > Right -> the same. Happy to double check with your POC :) > > > > > > > On Sat, Feb 22, 2020 at 6:16 PM Ash Berlin-Taylor > > > > > > > <a...@apache.org> > > > > > > > > > > > > > > > > > > wrote: > > > > > > > > > > > > > > > > That's a very good point about IDE's, I'll double check how > > > > > > > > PyCharm > > > > > > > behaves (Guessing PyCharm is the most popular one? PyCharm and > > > > > > > > > > > > > > > > > > > > > > > > > IntelliJ > > > > > > > > > > > > are > > > > > > > the same engine under the hood, right?) > > > > > > > > > > > > > > > > -a > > > > > > > > On Feb 22 2020, at 4:58 pm, Jarek Potiuk > > > > > > > > <jarek.pot...@polidea.com > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > wrote: > > > > > > > > > > My proposal here: > > > > > > > > > > Add in a _getattr_ based lazy import for DAG to > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > airflow/__init__.py > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > module > > > > > > > > > > > > > > > > > > > > > > > > > > > I am all for it - if we can do it in this way, then it is > > > > > > > > > indeed > > > > > > > > > better for the users. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > Do NOT issue a deprecation warning for this. > > > > > > > > > > Revert the change to all the imports in example dags etc so > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > that > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > those do from airflow import DAG. > > > > > > > > > I would only be for it if the PEP-562 change works fine with > > > > > > > > > IDE > > > > > > > > > support. Many of our users use IDEs to write their DAGsand > > > > > > > > > they > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > can > > > > > > > > > get confused where to import the DAGs from. If the PEP-562 > > > > > > > > > works > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > fine > > > > > > > > > with the IDEs, and they are smart enough to figure this out, I > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > am ok > > > > > > > > > with that. > > > > > > > > > > > > > > > > > > On a related note - I fully agree with the "clear API" > > > > > > > > > approach > > > > for > > > > > > > > > DAG developers. We did not have clear rules for that so far. > > > > > > > > > Eventually (I think 2.0 is good time for that) we should have > > > > > > > > > a > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > clear > > > > > > > > > statement what really the API is for DAG writers. We should > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > issue at > > > > > > > > > first a deprecation warning and then error if someone tries to > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > use > > > > > > > > > something not explicitly allowed from the DAG. It should not > > > > > > > > > be a > > > > > > > > > convention ("from airflow.* is allowed in documentation"). It > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > should > > > > > > > > > be enforcement. We can do it in scheduler - as we are parsing > > > > > > > > > the > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > DAGs > > > > > > > > > on our own and we can find out what imports the users are > > > > > > > > > using. > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > We > > > > > > > > > could even think about adding those deprecations in 2.0. There > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > are > > > > > > > > > > > > > > > > > > > > > > > > > > > not > > > > > > > > > that many things that should be importable from 'airflow.*'. > > > > > > > > > And > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > that > > > > > > > > > should be rather easy - I am playing now with AST parser and > > > > > > > > > it > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > is > > > > > > > > > rather straightforward - we just need to generate list of > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > "allowed" > > > > > > > > > imports . > > > > > > > > > > > > > > > > > > > (I'm less worried about the internal API airflow users > > > > internally, > > > > > > > so airflow.models.TaskInstance vs > > > > > > > > > > > > > > > > > > > > > > > > airflow.models.task_instance.TaskInstance > > > > > > > is okay for me. I find the later uglier but if it doesn't affect > > > > > > > > > > > > > > > > > > > > > > > > > users I > > > > > > > don't mind it) > > > > > > > > > > Ash > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > -- > > > > > > > > > Jarek Potiuk > > > > > > > > > Polidea | Principal Software Engineer > > > > > > > > > > > > > > > > > > M: +48 660 796 129 > > > > > > > -- > > > > > > > Jarek Potiuk > > > > > > > Polidea | Principal Software Engineer > > > > > > > > > > > > > > M: +48 660 796 129 > > > > > > -- > > > > > > Jarek Potiuk > > > > > > Polidea <https://www.polidea.com/> | Principal Software Engineer > > > > > > > > > > > > M: +48 660 796 129 <+48660796129> > > > > > > [image: Polidea] <https://www.polidea.com/> > > > > > > > > > > > > > > > >
