How to use inputs_to_dict ?

[jan-janssen]

I tried the following example:

from pyiron_workflow import Workflow, as_function_node
from pyiron_workflow.api import inputs_to_dict

@as_function_node
def return_dict(input_dict):
    return input_dict

wf = Workflow("test")
wf.a = 1
wf.input_dict_generator = inputs_to_dict(input_specification=["a", "b"])
wf.input_dict = wf.input_dict_generator(a=wf.a, b=2)
wf.result = return_dict(input_dict=wf.input_dict)
wf.draw(size=(10,10))

The input dict for result is not linked to the input_dict so I have the feeling I am doing something wrong:

[liamhuber]

You're extremely close. The inputs_to_dict:

pyiron_workflow/pyiron_workflow/nodes/transform.py

Line 253 in 66ab17b

Build a new :class:`InputsToDict` subclass and instantiate it.

However, given the lack of return hint and lack of docstring example, I still think you've raised a good and actionable issue. I'll improve the docstring later.

Concretely, your "generator" isn't a generator, but is just already the node you're looking for, so you can directly write:

from pyiron_workflow.api import Workflow, as_function_node, inputs_to_dict

@as_function_node
def return_dict(input_dict):
    return input_dict

wf = Workflow("test")
wf.a = 1
wf.input_dict = inputs_to_dict(input_specification=["a", "b"], a=wf.a, b=2)
wf.result = return_dict(input_dict=wf.input_dict)
wf.draw(size=(10,10))

[liamhuber]

Note that while I've jammed it all in one line, under the hood you're first building the node class with a and b and then populating those inputs:

pyiron_workflow/pyiron_workflow/nodes/transform.py

Lines 273 to 275 in 66ab17b

	cls = inputs_to_dict_factory(input_specification, class_name_suffix, use_cache)
	cls.preview_io()
	return cls(*node_args, **node_kwargs)

[jan-janssen]

Ok, that clarifies my issue. Thanks.