You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We currently store actions as a dict[str, Action] structure on Pipeline, where the str key is the name of the action. This makes some Action validation a little harder (and requires it's done on Pipeline instead of Action) since we want the name in the validator. We want to keep our existing YAML structure of:
actions:
action_name:
…
It's clear to users and we don't want to have to deal with keeping backwards compatibility.
So that leaves us with changing the structure of Actions inside the pipeline.
I've already tried mutating the dictionary coming out of parsed_yaml_file() but this led to having to change the dictionary nearly every test used which felt like a step backwards as the test data no longer looked similar to what a valid YAML file would look like.
The next step for this is to investigate if this can be done inside of the pydantic models and if so what niceties of pydantic are we opting out of, if any.
We currently store actions as a
dict[str, Action]
structure onPipeline
, where thestr
key is the name of the action. This makes some Action validation a little harder (and requires it's done on Pipeline instead of Action) since we want the name in the validator. We want to keep our existing YAML structure of:It's clear to users and we don't want to have to deal with keeping backwards compatibility.
So that leaves us with changing the structure of Actions inside the pipeline.
I've already tried mutating the dictionary coming out of
parsed_yaml_file()
but this led to having to change the dictionary nearly every test used which felt like a step backwards as the test data no longer looked similar to what a valid YAML file would look like.The next step for this is to investigate if this can be done inside of the pydantic models and if so what niceties of pydantic are we opting out of, if any.
Potentially #31 might be a route to take.
The text was updated successfully, but these errors were encountered: