Update invalid arg from snippet in TaskFlow tutorial doc

[josh-fell]

Doesn't really add anything to the documentation to substitute with a different arg either.
Updated from max_retries to retries and added to the following example in the doc since using retries when simply deserializing an output probably isn't a great use case to showcase using that arg. The next task tries to create a queue which seems like a more plausible place to use retries.

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in UPDATING.md.

[github-actions]

The PR is likely ready to be merged. No tests are needed as no important environment files, nor python files were modified by it. However, committers might decide that full test matrix is needed and add the 'full tests needed' label. Then you should rebase it to the latest main or amend the last commit of the PR, and push it with --force-with-lease.

[ashb]

I think the point here is to show that you can pass "operator" params to the task decorator, separately from the function arguments.

[josh-fell]

Agreed but I'm not sure it's adding anything in this section.

However, this has come up many times on Slack, StackOverflow, Discussions, etc., but I feel like a mini chapter on this in the tutorial is a good addition instead. (Something on my list I need to get to.) That way we can be really explicit about the functionality in TaskFlow.

Maybe even some TaskFlow "recipes" documentation.

[ashb]

This arg isn't invalid -- It's the rough equivalent of PythonOperator(max_retries=2, python_callable=x)

[josh-fell]

This arg isn't invalid -- It's the rough equivalent of PythonOperator(max_retries=2, python_callable=x)

Passing that specific arg causes DAG parsing errors. For example, given this DAG:

@task(max_retries=3)
def get_stuff(ti=None):
    print(vars(ti))


@dag(start_date=datetime(2022, 1, 1), schedule_interval=None)
def test_stuff():
    get_stuff()


dag = test_stuff()

You get this error:

[ashb]

Hmmm, I think that's a bug in the python decorator then!

[josh-fell]

I thought so too but I couldn't find "max_retries" anywhere that wasn't in the Amazon provider and/or pertinent to BaseOperator, PythonOperator, the decorators, and the like. Unless during my moonlighting I completely missed it.

retries, of course but not max_retries.

[josh-fell]

If it is a bug, I'd much rather fix that than these docs.

[potiuk]

I thought so too but I couldn't find "max_retries" anywhere that wasn't in the Amazon provider and/or pertinent to BaseOperator, PythonOperator, the decorators, and the like. Unless during my moonlighting I completely missed it.

I also have not found max_retries other than in providers, so I believe this was a mistake here.

[ashb]

🤦🏻 So yeah, this should just be @task(retries=3) then!

Still, the point being to show how you can pass baseoperator params on :)

[josh-fell]

I added retries to the next example. Didn't seem relevant to set a retry on a task that just deserializes an output. More likely to have retries creating a queue though which is what the next example shows.

[ashb]

Nice idea.

[josh-fell]

@ashb WDYT about the new variation on the update to the snippet?