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
A thing being empty might be an indication that the higher-level object that uses it should also be empty. For the use case that led me to develop stbl, I really need to allow specifically this!
to_cls(cls()) works, but to_cls_scalar(cls()) errors, even if allow_na = TRUE.
How should I handle this? Should I differentiate between empty and NULL? Should I sometimes coerce NULL to empty (particularly in stabilize_*)?
Judging from actual use, I think I probably want to default to allowing empty for to_cls_scalar(), but NOT allowing NULL.
The text was updated successfully, but these errors were encountered:
I'm leaning toward something along these lines: an empty arg (or something like that) replaces allow_null (everywhere?). It has possible values along the lines of: c("coerce_to_vec", "coerce_to_null", "accept", "reject").
A thing being empty might be an indication that the higher-level object that uses it should also be empty. For the use case that led me to develop stbl, I really need to allow specifically this!
to_cls(cls()) works, but to_cls_scalar(cls()) errors, even if allow_na = TRUE.
How should I handle this? Should I differentiate between empty and NULL? Should I sometimes coerce NULL to empty (particularly in stabilize_*)?
Judging from actual use, I think I probably want to default to allowing empty for to_cls_scalar(), but NOT allowing NULL.
The text was updated successfully, but these errors were encountered: