[Bug] Args not effective in direct ModelSpec construction about formulaic HOT 2 CLOSED

Hi @xjing76 ! Thanks for reporting.

There was a bug in formulaic 0.3.0 that would lead to this, but it should have been fixed in 0.3.1. Was it?

Your minimal example doesn't work for me (likely a different bug caused by the interaction of a structured formula with an unstructured model spec, looking into it now), but this code works fine against 0.3.1:

import pandas as pd
import formulaic
from formulaic.model_spec import ModelSpec
import scipy

data = [[1,10],[2,12],[1,13]]
df = pd.DataFrame(data,columns=['a','b'])
formula = 'b ~ 1+C(a)'
y, X_ = formulaic.Formula(formula).get_model_matrix(df.head(1))

kwargs = {'output':'sparse'}
spec = copy.deepcopy(X_.model_spec)
spec.output = 'sparse'
#spec = ModelSpec(formula=formula, structure=X_.model_spec.structure,  **kwargs)
X = spec.get_model_matrix(df)
X

<3x1 sparse matrix of type '<class 'numpy.int64'>'
	with 3 stored elements in Compressed Sparse Column format>

from formulaic.

Currently, ModelSpec's spanning left/right hand sides are not supported. ModelSpecs only describe single ModelMatrix objects. There are ways around this, but they're not elegant... I'll try to think about this for future releases.

This bug, however, does seem to be resolved, so I will close this out. Feel free to reopen of 0.3.1 doesn't work for you. I'll create a separate issue for the improvement of the user experience re-direct ModelSpec creation.

from formulaic.