Open truth-quark opened 5 months ago
amami
is supposed to be used only as a CLI program (manually and within scripts), not as a Python API.
amami
is supposed to be used only as a CLI program (manually and within scripts), not as a Python API.
Is there a particular reason for this?
What if someone wants to integrate um2nc
functionality into a workflow manager like airflow
or luigi
?
What if someone wants to integrate
um2nc
functionality into a workflow manager likeairflow
orluigi
?
I am not familiar with airflow
or luigi
workflow managers.
In any case, the purpose of amami
is to allow users to perform some defined tasks (such as converting a UM fieldsfile in netCDF with the um2nc
subcommand) that are useful (and not necessarily straightforward) within the climate community.
amami
uses some Python APIs to manipulate data internally (mainly xarray
, iris
and mule
).
If users want to perform more specific or complex tasks, they can use these APIs directly with their own scripts (potentially using any wokflow manager if they want).
For the moment I don't think we should support usage as an API. Then, in the future, if there is request for such a feature, we will implement it.
Is there any indication around all of the methods
amami
is going to be used in practice?Option #2 seems plausible if
um2nc
is called from a workflow tool or other code.