Closed ghisvail closed 3 months ago
Reproducing in https://github.com/nipype/pydra-nipype1/pull/35.
Should be fixed with pydra-nipype1 0.3.0.
Looks like the issue is resolved from the pydra-nipype1
side of things.
I am just wondering whether this part of the codebase should be made more robust? Is it logically sound to have both origin
and klass
set to None, or should we throw an exception with a more helpful message?
Maybe @tclose could give us his opinion on this use case?
Looks like the issue is resolved from the
pydra-nipype1
side of things.I am just wondering whether this part of the codebase should be made more robust? Is it logically sound to have both
origin
andklass
set to None, or should we throw an exception with a more helpful message?Maybe @tclose could give us his opinion on this use case?
So is the issue that None
should be effectively treated as NoneType
when used for typing? Seeing as though Python typing allows this, I'm ok with it in principle.
None
classes to be interpreted as NoneType
More issues discovered whilst trying out Pydra v0.23 with Clinica.
This time it's an issue when running a pipeline containing a Nipype1Task.
Which is fair, since:
So the first argument to
issubclass
is None, so indeed not a class.